class.media-summary.php — Jetpack – WP Security, Backup, Speed, & Growth 12.9.5

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

jetpack / _inc / lib Last commit date

class.media-summary.php

462 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	$return['type'] = 'video';
150	$return['video'] = esc_url_raw( 'http://www.youtube.com/watch?feature=player_embedded&v=' . $extract['shortcode']['youtube']['id'][0] );
151	$return['image'] = self::get_video_poster( 'youtube', $extract['shortcode']['youtube']['id'][0] );
152	$return['secure']['video'] = self::https( $return['video'] );
153	$return['secure']['image'] = self::https( $return['image'] );
154	}
155	++$return['count']['video'];
156	break;
157	case 'vimeo':
158	if ( 0 === $return['count']['video'] ) {
159	$return['type'] = 'video';
160	$return['video'] = esc_url_raw( 'http://vimeo.com/' . $extract['shortcode']['vimeo']['id'][0] );
161	$return['secure']['video'] = self::https( $return['video'] );
162
163	$poster_image = get_post_meta( $post_id, 'vimeo_poster_image', true );
164	if ( ! empty( $poster_image ) ) {
165	$return['image'] = $poster_image;
166	$poster_url_parts = wp_parse_url( $poster_image );
167	$return['secure']['image'] = 'https://secure-a.vimeocdn.com' . $poster_url_parts['path'];
168	}
169	}
170	++$return['count']['video'];
171	break;
172	}
173	}
174	}
175
176	if ( ! empty( $extract['has']['embed'] ) ) {
177	foreach ( $extract['embed']['url'] as $embed ) {
178	if ( preg_match( '/((youtube\|vimeo\|dailymotion)\.com\|youtu.be)/', $embed ) ) {
179	if ( 0 === $return['count']['video'] ) {
180	$return['type'] = 'video';
181	$return['video'] = 'http://' . $embed;
182	$return['secure']['video'] = self::https( $return['video'] );
183	if ( str_contains( $embed, 'youtube' ) ) {
184	$return['image'] = self::get_video_poster( 'youtube', jetpack_get_youtube_id( $return['video'] ) );
185	$return['secure']['image'] = self::https( $return['image'] );
186	} elseif ( str_contains( $embed, 'youtu.be' ) ) {
187	$youtube_id = jetpack_get_youtube_id( $return['video'] );
188	$return['video'] = 'http://youtube.com/watch?v=' . $youtube_id . '&feature=youtu.be';
189	$return['secure']['video'] = self::https( $return['video'] );
190	$return['image'] = self::get_video_poster( 'youtube', jetpack_get_youtube_id( $return['video'] ) );
191	$return['secure']['image'] = self::https( $return['image'] );
192	} elseif ( str_contains( $embed, 'vimeo' ) ) {
193	$poster_image = get_post_meta( $post_id, 'vimeo_poster_image', true );
194	if ( ! empty( $poster_image ) ) {
195	$return['image'] = $poster_image;
196	$poster_url_parts = wp_parse_url( $poster_image );
197	$return['secure']['image'] = 'https://secure-a.vimeocdn.com' . $poster_url_parts['path'];
198	}
199	} elseif ( str_contains( $embed, 'dailymotion' ) ) {
200	$return['image'] = str_replace( 'dailymotion.com/video/', 'dailymotion.com/thumbnail/video/', $embed );
201	$return['image'] = wp_parse_url( $return['image'], PHP_URL_SCHEME ) === null ? 'http://' . $return['image'] : $return['image'];
202	$return['secure']['image'] = self::https( $return['image'] );
203	}
204	}
205	++$return['count']['video'];
206	}
207	}
208	}
209
210	// Do we really want to make the video the primary focus of the post?
211	if ( 'video' === $return['type'] ) {
212	$content = wpautop( wp_strip_all_tags( $post->post_content ) );
213	$paragraphs = explode( '</p>', $content );
214	$number_of_paragraphs = 0;
215
216	foreach ( $paragraphs as $i => $paragraph ) {
217	// Don't include blank lines as a paragraph.
218	if ( '' === trim( $paragraph ) ) {
219	unset( $paragraphs[ $i ] );
220	continue;
221	}
222	++$number_of_paragraphs;
223	}
224
225	$number_of_paragraphs = $number_of_paragraphs - $return['count']['video']; // subtract amount for videos.
226
227	// More than 2 paragraph? The video is not the primary focus so we can do some more analysis.
228	if ( $number_of_paragraphs > 2 ) {
229	$return['type'] = 'standard';
230	}
231	}
232
233	// If we don't have any prioritized embed...
234	if ( 'standard' === $return['type'] ) {
235	if ( ( ! empty( $extract['has']['gallery'] ) \|\| ! empty( $extract['shortcode']['gallery']['count'] ) ) && ! empty( $extract['image'] ) ) {
236	// ... Then we prioritize galleries first (multiple images returned)
237	$return['type'] = 'gallery';
238	$return['images'] = $extract['image'];
239	foreach ( $return['images'] as $image ) {
240	$return['secure']['images'][] = array( 'url' => self::ssl_img( $image['url'] ) );
241	++$return['count']['image'];
242	}
243	} elseif ( ! empty( $extract['has']['image'] ) ) {
244	// ... Or we try and select a single image that would make sense.
245	$content = wpautop( wp_strip_all_tags( $post->post_content ) );
246	$paragraphs = explode( '</p>', $content );
247	$number_of_paragraphs = 0;
248
249	foreach ( $paragraphs as $i => $paragraph ) {
250	// Don't include 'actual' captions as a paragraph.
251	if ( str_contains( $paragraph, '[caption' ) ) {
252	unset( $paragraphs[ $i ] );
253	continue;
254	}
255	// Don't include blank lines as a paragraph.
256	if ( '' === trim( $paragraph ) ) {
257	unset( $paragraphs[ $i ] );
258	continue;
259	}
260	++$number_of_paragraphs;
261	}
262
263	$return['image'] = $extract['image'][0]['url'];
264	$return['secure']['image'] = self::ssl_img( $return['image'] );
265	++$return['count']['image'];
266
267	if ( $number_of_paragraphs <= 2 && is_countable( $extract['image'] ) && 1 === count( $extract['image'] ) ) {
268	// If we have lots of text or images, let's not treat it as an image post, but return its first image.
269	$return['type'] = 'image';
270	}
271	}
272	}
273
274	if ( $switched ) {
275	restore_current_blog();
276	}
277
278	/**
279	* Allow a theme or plugin to inspect and ultimately change the media summary.
280	*
281	* @since 4.4.0
282	*
283	* @param array $data The calculated media summary data.
284	* @param int $post_id The id of the post this data applies to.
285	*/
286	$return = apply_filters( 'jetpack_media_summary_output', $return, $post_id );
287
288	self::$cache[ $cache_key ] = $return;
289
290	return $return;
291	}
292
293	/**
294	* Converts http to https://
295	*
296	* @param string $str URL.
297	*
298	* @return string URL.
299	*/
300	public static function https( $str ) {
301	return str_replace( 'http://', 'https://', $str );
302	}
303
304	/**
305	* Returns a Photonized version of the URL.
306	*
307	* @param string $url URL.
308	*
309	* @return string URL.
310	*/
311	public static function ssl_img( $url ) {
312	if ( str_contains( $url, 'files.wordpress.com' ) ) {
313	return self::https( $url );
314	} else {
315	return self::https( Image_CDN_Core::cdn_url( $url ) );
316	}
317	}
318
319	/**
320	* Get the video poster.
321	*
322	* @param string $type Video service.
323	* @param string $id Video ID for the service.
324	*
325	* @return string URL of image thumbnail for the video.
326	*/
327	public static function get_video_poster( $type, $id ) {
328	if ( 'videopress' === $type ) {
329	if ( function_exists( 'video_get_highest_resolution_image_url' ) ) {
330	return video_get_highest_resolution_image_url( $id );
331	} elseif ( class_exists( 'VideoPress_Video' ) ) {
332	$video = new VideoPress_Video( $id );
333	return $video->poster_frame_uri;
334	}
335	} elseif ( 'youtube' === $type ) {
336	return 'http://img.youtube.com/vi/' . $id . '/0.jpg';
337	}
338	}
339
340	/**
341	* Clean text of shortcodes and tags.
342	*
343	* @param string $text Dirty text.
344	*
345	* @return string Clean text.
346	*/
347	public static function clean_text( $text ) {
348	return trim(
349	preg_replace(
350	'/[\s]+/',
351	' ',
352	preg_replace(
353	'@https?://[\S]+@',
354	'',
355	strip_shortcodes(
356	wp_strip_all_tags(
357	$text
358	)
359	)
360	)
361	)
362	);
363	}
364
365	/**
366	* Retrieve an excerpt for the post summary.
367	*
368	* This function works around a suspected problem with Core. If resolved, this function should be simplified.
369	*
370	* @link https://github.com/Automattic/jetpack/pull/8510
371	* @link https://core.trac.wordpress.org/ticket/42814
372	*
373	* @param string $post_content The post's content.
374	* @param string $post_excerpt The post's excerpt. Empty if none was explicitly set.
375	* @param int $max_words Maximum number of words for the excerpt. Used on wp.com. Default 16.
376	* @param int $max_chars Maximum characters in the excerpt. Used on wp.com. Default 256.
377	* @param WP_Post $requested_post The post object.
378	* @return string Post excerpt.
379	**/
380	public static function get_excerpt( $post_content, $post_excerpt, $max_words = 16, $max_chars = 256, $requested_post = null ) {
381	global $post;
382	$original_post = $post; // Saving the global for later use.
383	if ( empty( $post_excerpt ) && function_exists( 'wpcom_enhanced_excerpt_extract_excerpt' ) ) {
384	return self::clean_text(
385	wpcom_enhanced_excerpt_extract_excerpt(
386	array(
387	'text' => $post_content,
388	'excerpt_only' => true,
389	'show_read_more' => false,
390	'max_words' => $max_words,
391	'max_chars' => $max_chars,
392	'read_more_threshold' => 25,
393	)
394	)
395	);
396	} elseif ( $requested_post instanceof WP_Post ) {
397	// @todo Refactor to not need to override the global.
398	// phpcs:ignore: WordPress.WP.GlobalVariablesOverride.Prohibited
399	$post = $requested_post; // setup_postdata does not set the global.
400	setup_postdata( $post );
401	/** This filter is documented in core/src/wp-includes/post-template.php */
402	$post_excerpt = apply_filters( 'get_the_excerpt', $post_excerpt, $post );
403	// phpcs:ignore: WordPress.WP.GlobalVariablesOverride.Prohibited
404	$post = $original_post; // wp_reset_postdata uses the $post global.
405	wp_reset_postdata();
406	return self::clean_text( $post_excerpt );
407	}
408	return '';
409	}
410
411	/**
412	* Split a string into an array of words.
413	*
414	* @param string $text Post content or excerpt.
415	*
416	* @return array Array of words.
417	*/
418	public static function split_content_in_words( $text ) {
419	$words = preg_split( '/[\s!?;,.]+/', $text, -1, PREG_SPLIT_NO_EMPTY );
420
421	// Return an empty array if the split above fails.
422	return $words ? $words : array();
423	}
424
425	/**
426	* Get the word count.
427	*
428	* @param string $post_content Post content.
429	*
430	* @return int Word count.
431	*/
432	public static function get_word_count( $post_content ) {
433	return (int) count( self::split_content_in_words( self::clean_text( $post_content ) ) );
434	}
435
436	/**
437	* Get remainder word count (after the excerpt).
438	*
439	* @param string $post_content Post content.
440	* @param string $excerpt_content Excerpt content.
441	*
442	* @return int Number of words after the excerpt.
443	*/
444	public static function get_word_remaining_count( $post_content, $excerpt_content ) {
445	$content_word_count = count( self::split_content_in_words( self::clean_text( $post_content ) ) );
446	$excerpt_word_count = count( self::split_content_in_words( self::clean_text( $excerpt_content ) ) );
447
448	return (int) $content_word_count - $excerpt_word_count;
449	}
450
451	/**
452	* Counts the number of links in a post.
453	*
454	* @param string $post_content Post content.
455	*
456	* @return false\|int Number of links.
457	*/
458	public static function get_link_count( $post_content ) {
459	return preg_match_all( '/\<a[\> ]/', $post_content, $matches );
460	}
461	}
462