class.media-summary.php — Jetpack – WP Security, Backup, Speed, & Growth 13.7.2

jetpack / _inc / lib / class.media-summary.php

jetpack / _inc / lib Last commit date

class.media-summary.php

465 lines

1	<?php // phpcs:ignore WordPress.Files.FileName.InvalidClassFileName
2	/**
3	* Provides media summary of a post.
4	*
5	* @package automattic/jetpack
6	*/
7
8	use Automattic\Jetpack\Image_CDN\Image_CDN_Core;
9
10	/**
11	* Class Jetpack_Media_Summary
12	*
13	* Priority: embed [video] > gallery > image > text
14	*/
15	class Jetpack_Media_Summary {
16
17	/**
18	* Media cache.
19	*
20	* @var array
21	*/
22	private static $cache = array();
23
24	/**
25	* Get media summary for a post.
26	*
27	* @param int $post_id Post ID.
28	* @param int $blog_id Blog ID, if applicable.
29	* @param array $args {
30	* Optional. An array of arguments.
31	* @type int $max_words Maximum number of words.
32	* @type int $max_chars Maximum number of characters.
33	* }
34	*
35	* @return array\|mixed\|void
36	*/
37	public static function get( $post_id, $blog_id = 0, $args = array() ) {
38	// @todo: Use type hinting in the line above when at PHP 7.0+.
39	$post_id = (int) $post_id;
40	$blog_id = (int) $blog_id;
41
42	$defaults = array(
43	'max_words' => 16,
44	'max_chars' => 256,
45	);
46	$args = wp_parse_args( $args, $defaults );
47
48	$switched = false;
49	if ( ! empty( $blog_id ) && get_current_blog_id() !== $blog_id && function_exists( 'switch_to_blog' ) ) {
50	switch_to_blog( $blog_id );
51	$switched = true;
52	} else {
53	$blog_id = get_current_blog_id();
54	}
55
56	$cache_key = "{$blog_id}_{$post_id}_{$args['max_words']}_{$args['max_chars']}";
57	if ( isset( self::$cache[ $cache_key ] ) ) {
58	if ( $switched ) {
59	restore_current_blog();
60	}
61	return self::$cache[ $cache_key ];
62	}
63
64	if ( ! class_exists( 'Jetpack_Media_Meta_Extractor' ) ) {
65	require_once JETPACK__PLUGIN_DIR . '_inc/lib/class.media-extractor.php';
66	}
67
68	$post = get_post( $post_id );
69	$permalink = get_permalink( $post_id );
70
71	$return = array(
72	'type' => 'standard',
73	'permalink' => $permalink,
74	'image' => '',
75	'excerpt' => '',
76	'word_count' => 0,
77	'secure' => array(
78	'image' => '',
79	),
80	'count' => array(
81	'image' => 0,
82	'video' => 0,
83	'word' => 0,
84	'link' => 0,
85	),
86	);
87
88	if ( $post instanceof WP_Post && empty( $post->post_password ) ) {
89	$return['excerpt'] = self::get_excerpt( $post->post_content, $post->post_excerpt, $args['max_words'], $args['max_chars'], $post );
90	$return['count']['word'] = self::get_word_count( $post->post_content );
91	$return['count']['word_remaining'] = self::get_word_remaining_count( $post->post_content, $return['excerpt'] );
92	$return['count']['link'] = self::get_link_count( $post->post_content );
93	}
94
95	$extract = Jetpack_Media_Meta_Extractor::extract( $blog_id, $post_id, Jetpack_Media_Meta_Extractor::ALL );
96
97	if ( empty( $extract['has'] ) ) {
98	if ( $switched ) {
99	restore_current_blog();
100	}
101	self::$cache[ $cache_key ] = $return;
102	return $return;
103	}
104
105	// Prioritize [some] video embeds.
106	if ( ! empty( $extract['has']['shortcode'] ) ) {
107	foreach ( $extract['shortcode'] as $type => $data ) {
108	switch ( $type ) {
109	case 'videopress':
110	case 'wpvideo':
111	if ( 0 === $return['count']['video'] ) {
112	// If there is no id on the video, then let's just skip this.
113	if ( ! isset( $data['id'][0] ) ) {
114	break;
115	}
116
117	$guid = $data['id'][0];
118	$video_info = videopress_get_video_details( $guid );
119
120	// Only add the video tags if the guid returns a valid videopress object.
121	if ( $video_info instanceof stdClass ) {
122	// Continue early if we can't find a Video slug.
123	if ( empty( $video_info->files->std->mp4 ) ) {
124	break;
125	}
126
127	$url = sprintf(
128	'https://videos.files.wordpress.com/%1$s/%2$s',
129	$guid,
130	$video_info->files->std->mp4
131	);
132
133	$thumbnail = $video_info->poster;
134	if ( ! empty( $thumbnail ) ) {
135	$return['image'] = $thumbnail;
136	$return['secure']['image'] = $thumbnail;
137	}
138
139	$return['type'] = 'video';
140	$return['video'] = esc_url_raw( $url );
141	$return['video_type'] = 'video/mp4';
142	$return['secure']['video'] = $return['video'];
143	}
144	}
145	++$return['count']['video'];
146	break;
147	case 'youtube':
148	if ( 0 === $return['count']['video'] ) {
149	if ( ! isset( $extract['shortcode']['youtube']['id'][0] ) ) {
150	break;
151	}
152	$return['type'] = 'video';
153	$return['video'] = esc_url_raw( 'http://www.youtube.com/watch?feature=player_embedded&v=' . $extract['shortcode']['youtube']['id'][0] );
154	$return['image'] = self::get_video_poster( 'youtube', $extract['shortcode']['youtube']['id'][0] );
155	$return['secure']['video'] = self::https( $return['video'] );
156	$return['secure']['image'] = self::https( $return['image'] );
157	}
158	++$return['count']['video'];
159	break;
160	case 'vimeo':
161	if ( 0 === $return['count']['video'] ) {
162	$return['type'] = 'video';
163	$return['video'] = esc_url_raw( 'http://vimeo.com/' . $extract['shortcode']['vimeo']['id'][0] );
164	$return['secure']['video'] = self::https( $return['video'] );
165
166	$poster_image = get_post_meta( $post_id, 'vimeo_poster_image', true );
167	if ( ! empty( $poster_image ) ) {
168	$return['image'] = $poster_image;
169	$poster_url_parts = wp_parse_url( $poster_image );
170	$return['secure']['image'] = 'https://secure-a.vimeocdn.com' . $poster_url_parts['path'];
171	}
172	}
173	++$return['count']['video'];
174	break;
175	}
176	}
177	}
178
179	if ( ! empty( $extract['has']['embed'] ) ) {
180	foreach ( $extract['embed']['url'] as $embed ) {
181	if ( preg_match( '/((youtube\|vimeo\|dailymotion)\.com\|youtu.be)/', $embed ) ) {
182	if ( 0 === $return['count']['video'] ) {
183	$return['type'] = 'video';
184	$return['video'] = 'http://' . $embed;
185	$return['secure']['video'] = self::https( $return['video'] );
186	if ( str_contains( $embed, 'youtube' ) ) {
187	$return['image'] = self::get_video_poster( 'youtube', jetpack_get_youtube_id( $return['video'] ) );
188	$return['secure']['image'] = self::https( $return['image'] );
189	} elseif ( str_contains( $embed, 'youtu.be' ) ) {
190	$youtube_id = jetpack_get_youtube_id( $return['video'] );
191	$return['video'] = 'http://youtube.com/watch?v=' . $youtube_id . '&feature=youtu.be';
192	$return['secure']['video'] = self::https( $return['video'] );
193	$return['image'] = self::get_video_poster( 'youtube', jetpack_get_youtube_id( $return['video'] ) );
194	$return['secure']['image'] = self::https( $return['image'] );
195	} elseif ( str_contains( $embed, 'vimeo' ) ) {
196	$poster_image = get_post_meta( $post_id, 'vimeo_poster_image', true );
197	if ( ! empty( $poster_image ) ) {
198	$return['image'] = $poster_image;
199	$poster_url_parts = wp_parse_url( $poster_image );
200	$return['secure']['image'] = 'https://secure-a.vimeocdn.com' . $poster_url_parts['path'];
201	}
202	} elseif ( str_contains( $embed, 'dailymotion' ) ) {
203	$return['image'] = str_replace( 'dailymotion.com/video/', 'dailymotion.com/thumbnail/video/', $embed );
204	$return['image'] = wp_parse_url( $return['image'], PHP_URL_SCHEME ) === null ? 'http://' . $return['image'] : $return['image'];
205	$return['secure']['image'] = self::https( $return['image'] );
206	}
207	}
208	++$return['count']['video'];
209	}
210	}
211	}
212
213	// Do we really want to make the video the primary focus of the post?
214	if ( 'video' === $return['type'] ) {
215	$content = wpautop( wp_strip_all_tags( $post->post_content ) );
216	$paragraphs = explode( '</p>', $content );
217	$number_of_paragraphs = 0;
218
219	foreach ( $paragraphs as $i => $paragraph ) {
220	// Don't include blank lines as a paragraph.
221	if ( '' === trim( $paragraph ) ) {
222	unset( $paragraphs[ $i ] );
223	continue;
224	}
225	++$number_of_paragraphs;
226	}
227
228	$number_of_paragraphs = $number_of_paragraphs - $return['count']['video']; // subtract amount for videos.
229
230	// More than 2 paragraph? The video is not the primary focus so we can do some more analysis.
231	if ( $number_of_paragraphs > 2 ) {
232	$return['type'] = 'standard';
233	}
234	}
235
236	// If we don't have any prioritized embed...
237	if ( 'standard' === $return['type'] ) {
238	if ( ( ! empty( $extract['has']['gallery'] ) \|\| ! empty( $extract['shortcode']['gallery']['count'] ) ) && ! empty( $extract['image'] ) ) {
239	// ... Then we prioritize galleries first (multiple images returned)
240	$return['type'] = 'gallery';
241	$return['images'] = $extract['image'];
242	foreach ( $return['images'] as $image ) {
243	$return['secure']['images'][] = array( 'url' => self::ssl_img( $image['url'] ) );
244	++$return['count']['image'];
245	}
246	} elseif ( ! empty( $extract['has']['image'] ) ) {
247	// ... Or we try and select a single image that would make sense.
248	$content = wpautop( wp_strip_all_tags( $post->post_content ) );
249	$paragraphs = explode( '</p>', $content );
250	$number_of_paragraphs = 0;
251
252	foreach ( $paragraphs as $i => $paragraph ) {
253	// Don't include 'actual' captions as a paragraph.
254	if ( str_contains( $paragraph, '[caption' ) ) {
255	unset( $paragraphs[ $i ] );
256	continue;
257	}
258	// Don't include blank lines as a paragraph.
259	if ( '' === trim( $paragraph ) ) {
260	unset( $paragraphs[ $i ] );
261	continue;
262	}
263	++$number_of_paragraphs;
264	}
265
266	$return['image'] = $extract['image'][0]['url'];
267	$return['secure']['image'] = self::ssl_img( $return['image'] );
268	++$return['count']['image'];
269
270	if ( $number_of_paragraphs <= 2 && is_countable( $extract['image'] ) && 1 === count( $extract['image'] ) ) {
271	// If we have lots of text or images, let's not treat it as an image post, but return its first image.
272	$return['type'] = 'image';
273	}
274	}
275	}
276
277	if ( $switched ) {
278	restore_current_blog();
279	}
280
281	/**
282	* Allow a theme or plugin to inspect and ultimately change the media summary.
283	*
284	* @since 4.4.0
285	*
286	* @param array $data The calculated media summary data.
287	* @param int $post_id The id of the post this data applies to.
288	*/
289	$return = apply_filters( 'jetpack_media_summary_output', $return, $post_id );
290
291	self::$cache[ $cache_key ] = $return;
292
293	return $return;
294	}
295
296	/**
297	* Converts http to https://
298	*
299	* @param string $str URL.
300	*
301	* @return string URL.
302	*/
303	public static function https( $str ) {
304	return str_replace( 'http://', 'https://', $str );
305	}
306
307	/**
308	* Returns a Photonized version of the URL.
309	*
310	* @param string $url URL.
311	*
312	* @return string URL.
313	*/
314	public static function ssl_img( $url ) {
315	if ( str_contains( $url, 'files.wordpress.com' ) ) {
316	return self::https( $url );
317	} else {
318	return self::https( Image_CDN_Core::cdn_url( $url ) );
319	}
320	}
321
322	/**
323	* Get the video poster.
324	*
325	* @param string $type Video service.
326	* @param string $id Video ID for the service.
327	*
328	* @return string URL of image thumbnail for the video.
329	*/
330	public static function get_video_poster( $type, $id ) {
331	if ( 'videopress' === $type ) {
332	if ( function_exists( 'video_get_highest_resolution_image_url' ) ) {
333	return video_get_highest_resolution_image_url( $id );
334	} elseif ( class_exists( 'VideoPress_Video' ) ) {
335	$video = new VideoPress_Video( $id );
336	return $video->poster_frame_uri;
337	}
338	} elseif ( 'youtube' === $type ) {
339	return 'http://img.youtube.com/vi/' . $id . '/0.jpg';
340	}
341	}
342
343	/**
344	* Clean text of shortcodes and tags.
345	*
346	* @param string $text Dirty text.
347	*
348	* @return string Clean text.
349	*/
350	public static function clean_text( $text ) {
351	return trim(
352	preg_replace(
353	'/[\s]+/',
354	' ',
355	preg_replace(
356	'@https?://[\S]+@',
357	'',
358	strip_shortcodes(
359	wp_strip_all_tags(
360	$text
361	)
362	)
363	)
364	)
365	);
366	}
367
368	/**
369	* Retrieve an excerpt for the post summary.
370	*
371	* This function works around a suspected problem with Core. If resolved, this function should be simplified.
372	*
373	* @link https://github.com/Automattic/jetpack/pull/8510
374	* @link https://core.trac.wordpress.org/ticket/42814
375	*
376	* @param string $post_content The post's content.
377	* @param string $post_excerpt The post's excerpt. Empty if none was explicitly set.
378	* @param int $max_words Maximum number of words for the excerpt. Used on wp.com. Default 16.
379	* @param int $max_chars Maximum characters in the excerpt. Used on wp.com. Default 256.
380	* @param WP_Post $requested_post The post object.
381	* @return string Post excerpt.
382	**/
383	public static function get_excerpt( $post_content, $post_excerpt, $max_words = 16, $max_chars = 256, $requested_post = null ) {
384	global $post;
385	$original_post = $post; // Saving the global for later use.
386	if ( empty( $post_excerpt ) && function_exists( 'wpcom_enhanced_excerpt_extract_excerpt' ) ) {
387	return self::clean_text(
388	wpcom_enhanced_excerpt_extract_excerpt(
389	array(
390	'text' => $post_content,
391	'excerpt_only' => true,
392	'show_read_more' => false,
393	'max_words' => $max_words,
394	'max_chars' => $max_chars,
395	'read_more_threshold' => 25,
396	)
397	)
398	);
399	} elseif ( $requested_post instanceof WP_Post ) {
400	// @todo Refactor to not need to override the global.
401	// phpcs:ignore: WordPress.WP.GlobalVariablesOverride.Prohibited
402	$post = $requested_post; // setup_postdata does not set the global.
403	setup_postdata( $post );
404	/** This filter is documented in core/src/wp-includes/post-template.php */
405	$post_excerpt = apply_filters( 'get_the_excerpt', $post_excerpt, $post );
406	// phpcs:ignore: WordPress.WP.GlobalVariablesOverride.Prohibited
407	$post = $original_post; // wp_reset_postdata uses the $post global.
408	wp_reset_postdata();
409	return self::clean_text( $post_excerpt );
410	}
411	return '';
412	}
413
414	/**
415	* Split a string into an array of words.
416	*
417	* @param string $text Post content or excerpt.
418	*
419	* @return array Array of words.
420	*/
421	public static function split_content_in_words( $text ) {
422	$words = preg_split( '/[\s!?;,.]+/', $text, -1, PREG_SPLIT_NO_EMPTY );
423
424	// Return an empty array if the split above fails.
425	return $words ? $words : array();
426	}
427
428	/**
429	* Get the word count.
430	*
431	* @param string $post_content Post content.
432	*
433	* @return int Word count.
434	*/
435	public static function get_word_count( $post_content ) {
436	return (int) count( self::split_content_in_words( self::clean_text( $post_content ) ) );
437	}
438
439	/**
440	* Get remainder word count (after the excerpt).
441	*
442	* @param string $post_content Post content.
443	* @param string $excerpt_content Excerpt content.
444	*
445	* @return int Number of words after the excerpt.
446	*/
447	public static function get_word_remaining_count( $post_content, $excerpt_content ) {
448	$content_word_count = count( self::split_content_in_words( self::clean_text( $post_content ) ) );
449	$excerpt_word_count = count( self::split_content_in_words( self::clean_text( $excerpt_content ) ) );
450
451	return (int) $content_word_count - $excerpt_word_count;
452	}
453
454	/**
455	* Counts the number of links in a post.
456	*
457	* @param string $post_content Post content.
458	*
459	* @return false\|int Number of links.
460	*/
461	public static function get_link_count( $post_content ) {
462	return preg_match_all( '/\<a[\> ]/', $post_content, $matches );
463	}
464	}
465