wordpress-seo / inc / sitemaps / class-sitemap-image-parser.php

class-sitemap-image-parser.php in Yoast SEO – Advanced SEO with real-time guidance and built-in AI 28.2, at inc/sitemaps/class-sitemap-image-parser.php

510 lines 11.7 KB

Raw Download Zip

1	<?php
2	/**
3	* WPSEO plugin file.
4	*
5	* @package WPSEO\XML_Sitemaps
6	*/
7
8	/**
9	* Parses images from the given post.
10	*/
11	class WPSEO_Sitemap_Image_Parser {
12
13	/**
14	* Holds the home_url() value to speed up loops.
15	*
16	* @var string
17	*/
18	protected $home_url = '';
19
20	/**
21	* Holds site URL hostname.
22	*
23	* @var string
24	*/
25	protected $host = '';
26
27	/**
28	* Holds site URL protocol.
29	*
30	* @var string
31	*/
32	protected $scheme = 'http';
33
34	/**
35	* Cached set of attachments for multiple posts.
36	*
37	* @var array
38	*/
39	protected $attachments = [];
40
41	/**
42	* Holds blog charset value for use in DOM parsing.
43	*
44	* @var string
45	*/
46	protected $charset = 'UTF-8';
47
48	/**
49	* Set up URL properties for reuse.
50	*/
51	public function __construct() {
52
53	$this->home_url = home_url();
54	$parsed_home = wp_parse_url( $this->home_url );
55
56	if ( ! empty( $parsed_home['host'] ) ) {
57	$this->host = str_replace( 'www.', '', $parsed_home['host'] );
58	}
59
60	if ( ! empty( $parsed_home['scheme'] ) ) {
61	$this->scheme = $parsed_home['scheme'];
62	}
63
64	$this->charset = esc_attr( get_bloginfo( 'charset' ) );
65	}
66
67	/**
68	* Get set of image data sets for the given post.
69	*
70	* @param object $post Post object to get images for.
71	*
72	* @return array
73	*/
74	public function get_images( $post ) {
75
76	$images = [];
77
78	if ( ! is_object( $post ) ) {
79	return $images;
80	}
81
82	$thumbnail_id = get_post_thumbnail_id( $post->ID );
83
84	if ( $thumbnail_id ) {
85
86	$src = $this->get_absolute_url( $this->image_url( $thumbnail_id ) );
87	$images[] = $this->get_image_item( $post, $src );
88	}
89
90	/**
91	* Filter: 'wpseo_sitemap_content_before_parse_html_images' - Filters the post content
92	* before it is parsed for images.
93	*
94	* @param string $content The raw/unprocessed post content.
95	*/
96	$content = apply_filters( 'wpseo_sitemap_content_before_parse_html_images', $post->post_content );
97
98	$unfiltered_images = $this->parse_html_images( $content );
99
100	foreach ( $unfiltered_images as $image ) {
101	$images[] = $this->get_image_item( $post, $image['src'] );
102	}
103
104	foreach ( $this->parse_galleries( $content, $post->ID ) as $attachment ) {
105	$src = $this->get_absolute_url( $this->image_url( $attachment->ID ) );
106	$images[] = $this->get_image_item( $post, $src );
107	}
108
109	if ( $post->post_type === 'attachment' && wp_attachment_is_image( $post ) ) {
110	$src = $this->get_absolute_url( $this->image_url( $post->ID ) );
111	$images[] = $this->get_image_item( $post, $src );
112	}
113
114	foreach ( $images as $key => $image ) {
115
116	if ( empty( $image['src'] ) ) {
117	unset( $images[ $key ] );
118	}
119	}
120
121	/**
122	* Filter images to be included for the post in XML sitemap.
123	*
124	* @param array $images Array of image items.
125	* @param int $post_id ID of the post.
126	*/
127	$image_list = apply_filters( 'wpseo_sitemap_urlimages', $images, $post->ID );
128	if ( isset( $image_list ) && is_array( $image_list ) ) {
129	$images = $image_list;
130	}
131
132	return $images;
133	}
134
135	/**
136	* Get the images in the term description.
137	*
138	* @param object $term Term to get images from description for.
139	*
140	* @return array
141	*/
142	public function get_term_images( $term ) {
143
144	$images = $this->parse_html_images( $term->description );
145
146	foreach ( $this->parse_galleries( $term->description ) as $attachment ) {
147
148	$images[] = [
149	'src' => $this->get_absolute_url( $this->image_url( $attachment->ID ) ),
150	];
151	}
152
153	/**
154	* Filter images to be included for the term in XML sitemap.
155	*
156	* @param array $image_list Array of image items.
157	* @param int $term_id ID of the post.
158	*/
159	$image_list = apply_filters( 'wpseo_sitemap_urlimages_term', $images, $term->term_id );
160	if ( isset( $image_list ) && is_array( $image_list ) ) {
161	$images = $image_list;
162	}
163
164	return $images;
165	}
166
167	/**
168	* Parse `<img />` tags in content.
169	*
170	* @param string $content Content string to parse.
171	*
172	* @return array
173	*/
174	private function parse_html_images( $content ) {
175
176	$images = [];
177
178	if ( ! class_exists( 'DOMDocument' ) ) {
179	return $images;
180	}
181
182	if ( empty( $content ) ) {
183	return $images;
184	}
185
186	// Prevent DOMDocument from bubbling warnings about invalid HTML.
187	libxml_use_internal_errors( true );
188
189	$post_dom = new DOMDocument();
190	$post_dom->loadHTML( '<?xml encoding="' . $this->charset . '">' . $content );
191
192	// Clear the errors, so they don't get kept in memory.
193	libxml_clear_errors();
194
195	/**
196	* Image attribute.
197	*
198	* @var DOMElement $img
199	*/
200	foreach ( $post_dom->getElementsByTagName( 'img' ) as $img ) {
201
202	$src = $img->getAttribute( 'src' );
203
204	if ( empty( $src ) ) {
205	continue;
206	}
207
208	$class = $img->getAttribute( 'class' );
209
210	if ( // This detects WP-inserted images, which we need to upsize. R.
211	! empty( $class )
212	&& ( strpos( $class, 'size-full' ) === false )
213	&& preg_match( '\|wp-image-(?P<id>\d+)\|', $class, $matches )
214	&& get_post_status( $matches['id'] )
215	) {
216	$query_params = wp_parse_url( $src, PHP_URL_QUERY );
217	$src = $this->image_url( $matches['id'] );
218
219	if ( $query_params ) {
220	$src .= '?' . $query_params;
221	}
222	}
223
224	$src = $this->get_absolute_url( $src );
225
226	if ( strpos( $src, $this->host ) === false ) {
227	continue;
228	}
229
230	if ( $src !== esc_url( $src, null, 'attribute' ) ) {
231	continue;
232	}
233
234	$images[] = [
235	'src' => $src,
236	];
237	}
238
239	return $images;
240	}
241
242	/**
243	* Parse gallery shortcodes in a given content.
244	*
245	* @param string $content Content string.
246	* @param int $post_id Optional. ID of post being parsed.
247	*
248	* @return array Set of attachment objects.
249	*/
250	protected function parse_galleries( $content, $post_id = 0 ) {
251
252	$attachments = [];
253	$galleries = $this->get_content_galleries( $content );
254
255	foreach ( $galleries as $gallery ) {
256
257	$id = $post_id;
258
259	if ( ! empty( $gallery['id'] ) ) {
260	$id = (int) $gallery['id'];
261	}
262
263	// Forked from core gallery_shortcode() to have exact same logic. R.
264	if ( ! empty( $gallery['ids'] ) ) {
265	$gallery['include'] = $gallery['ids'];
266	}
267
268	$gallery_attachments = $this->get_gallery_attachments( $id, $gallery );
269
270	$attachments = array_merge( $attachments, $gallery_attachments );
271	}
272
273	return array_unique( $attachments, SORT_REGULAR );
274	}
275
276	/**
277	* Retrieves galleries from the passed content.
278	*
279	* Forked from core to skip executing shortcodes for performance.
280	*
281	* @param string $content Content to parse for shortcodes.
282	*
283	* @return array A list of arrays, each containing gallery data.
284	*/
285	protected function get_content_galleries( $content ) {
286
287	$galleries = [];
288
289	if ( ! preg_match_all( '/' . get_shortcode_regex( [ 'gallery' ] ) . '/s', $content, $matches, PREG_SET_ORDER ) ) {
290	return $galleries;
291	}
292
293	foreach ( $matches as $shortcode ) {
294
295	$attributes = shortcode_parse_atts( $shortcode[3] );
296
297	if ( $attributes === '' ) { // Valid shortcode without any attributes. R.
298	$attributes = [];
299	}
300
301	$galleries[] = $attributes;
302	}
303
304	return $galleries;
305	}
306
307	/**
308	* Get image item array with filters applied.
309	*
310	* @param WP_Post $post Post object for the context.
311	* @param string $src Image URL.
312	*
313	* @return array
314	*/
315	protected function get_image_item( $post, $src ) {
316
317	$image = [];
318
319	/**
320	* Filter image URL to be included in XML sitemap for the post.
321	*
322	* @param string $src Image URL.
323	* @param object $post Post object.
324	*/
325	$image['src'] = apply_filters( 'wpseo_xml_sitemap_img_src', $src, $post );
326
327	/**
328	* Filter image data to be included in XML sitemap for the post.
329	*
330	* @param array $image {
331	* Array of image data.
332	*
333	* @type string $src Image URL.
334	* }
335	*
336	* @param object $post Post object.
337	*/
338	return apply_filters( 'wpseo_xml_sitemap_img', $image, $post );
339	}
340
341	/**
342	* Get attached image URL with filters applied. Adapted from core for speed.
343	*
344	* @param int $post_id ID of the post.
345	*
346	* @return string
347	*/
348	private function image_url( $post_id ) {
349
350	static $uploads;
351
352	if ( empty( $uploads ) ) {
353	$uploads = wp_upload_dir();
354	}
355
356	if ( $uploads['error'] !== false ) {
357	return '';
358	}
359
360	$file = get_post_meta( $post_id, '_wp_attached_file', true );
361
362	if ( empty( $file ) ) {
363	return '';
364	}
365
366	// Check that the upload base exists in the file location.
367	if ( strpos( $file, $uploads['basedir'] ) === 0 ) {
368	$src = str_replace( $uploads['basedir'], $uploads['baseurl'], $file );
369	}
370	elseif ( strpos( $file, 'wp-content/uploads' ) !== false ) {
371	$src = $uploads['baseurl'] . substr( $file, ( strpos( $file, 'wp-content/uploads' ) + 18 ) );
372	}
373	else {
374	// It's a newly uploaded file, therefore $file is relative to the baseurl.
375	$src = $uploads['baseurl'] . '/' . $file;
376	}
377
378	return apply_filters( 'wp_get_attachment_url', $src, $post_id );
379	}
380
381	/**
382	* Make absolute URL for domain or protocol-relative one.
383	*
384	* @param string $src URL to process.
385	*
386	* @return string
387	*/
388	protected function get_absolute_url( $src ) {
389
390	if ( empty( $src ) \|\| ! is_string( $src ) ) {
391	return $src;
392	}
393
394	if ( YoastSEO()->helpers->url->is_relative( $src ) === true ) {
395
396	if ( $src[0] !== '/' ) {
397	return $src;
398	}
399
400	// The URL is relative, we'll have to make it absolute.
401	return $this->home_url . $src;
402	}
403
404	if ( strpos( $src, 'http' ) !== 0 ) {
405	// Protocol relative URL, we add the scheme as the standard requires a protocol.
406	return $this->scheme . ':' . $src;
407	}
408
409	return $src;
410	}
411
412	/**
413	* Returns the attachments for a gallery.
414	*
415	* @param int $id The post ID.
416	* @param array $gallery The gallery config.
417	*
418	* @return array The selected attachments.
419	*/
420	protected function get_gallery_attachments( $id, $gallery ) {
421
422	// When there are attachments to include.
423	if ( ! empty( $gallery['include'] ) ) {
424	return $this->get_gallery_attachments_for_included( $gallery['include'] );
425	}
426
427	// When $id is empty, just return empty array.
428	if ( empty( $id ) ) {
429	return [];
430	}
431
432	return $this->get_gallery_attachments_for_parent( $id, $gallery );
433	}
434
435	/**
436	* Returns the attachments for the given ID.
437	*
438	* @param int $id The post ID.
439	* @param array $gallery The gallery config.
440	*
441	* @return array The selected attachments.
442	*/
443	protected function get_gallery_attachments_for_parent( $id, $gallery ) {
444	$query = [
445	'posts_per_page' => -1,
446	'post_parent' => $id,
447	];
448
449	// When there are posts that should be excluded from result set.
450	if ( ! empty( $gallery['exclude'] ) ) {
451	$query['post__not_in'] = wp_parse_id_list( $gallery['exclude'] );
452	}
453
454	return $this->get_attachments( $query );
455	}
456
457	/**
458	* Returns an array with attachments for the post IDs that will be included.
459	*
460	* @param array $included_ids Array with IDs to include.
461	*
462	* @return array The found attachments.
463	*/
464	protected function get_gallery_attachments_for_included( $included_ids ) {
465	$ids_to_include = wp_parse_id_list( $included_ids );
466	$attachments = $this->get_attachments(
467	[
468	'posts_per_page' => count( $ids_to_include ),
469	'post__in' => $ids_to_include,
470	],
471	);
472
473	$gallery_attachments = [];
474	foreach ( $attachments as $val ) {
475	$gallery_attachments[ $val->ID ] = $val;
476	}
477
478	return $gallery_attachments;
479	}
480
481	/**
482	* Returns the attachments.
483	*
484	* @param array $args Array with query args.
485	*
486	* @return array The found attachments.
487	*/
488	protected function get_attachments( $args ) {
489	$default_args = [
490	'post_status' => 'inherit',
491	'post_type' => 'attachment',
492	'post_mime_type' => 'image',
493
494	// Defaults taken from function get_posts.
495	'orderby' => 'date',
496	'order' => 'DESC',
497	'meta_key' => '',
498	'meta_value' => '',
499	'suppress_filters' => true,
500	'ignore_sticky_posts' => true,
501	'no_found_rows' => true,
502	];
503
504	$args = wp_parse_args( $args, $default_args );
505
506	$get_attachments = new WP_Query();
507	return $get_attachments->query( $args );
508	}
509	}
510