class.media-summary.php — Jetpack – WP Security, Backup, Speed, & Growth 12.0.3

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

jetpack / _inc / lib Last commit date

class.media-summary.php

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