speechkit / src / Component / Post / PostContentUtils.php

PostContentUtils.php in BeyondWords – AI audio for publishers 4.1.1, at src/Component/Post/PostContentUtils.php

465 lines 14.8 KB

Raw Download Zip

1	<?php
2
3	declare(strict_types=1);
4
5	namespace Beyondwords\Wordpress\Component\Post;
6
7	/**
8	* BeyondWords Post Content Utilities.
9	*
10	* @package Beyondwords
11	* @subpackage Beyondwords/includes
12	* @author Stuart McAlpine <stu@beyondwords.io>
13	* @since 3.5.0
14	*/
15	class PostContentUtils
16	{
17	public const DATE_FORMAT = 'Y-m-d\TH:i:s\Z';
18
19	/**
20	* Get the source text for the audio, ready to be sent to the BeyondWords API.
21	*
22	* @deprecated 4.0.0 Renamed to PostContentUtils::getBody()
23	*
24	* @param int\|WP_Post $post The WordPress post ID, or post object.
25	*
26	* @since 3.0.0
27	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
28	* @since 3.8.0 Exclude Gutenberg blocks with attribute { beyondwordsAudio: false }
29	* @since 4.0.0 Renamed from PostContentUtils::getSourceTextForAudio() to PostContentUtils::getBody()
30	*
31	* @return string The body (the processed $post->post_content).
32	*/
33	public static function getSourceTextForAudio($post)
34	{
35	_doing_it_wrong(
36	'PostContentUtils::getBody',
37	'BeyondWords PostContentUtils::getSourceTextForAudio() has been renamed to PostContentUtils::getBody()',
38	'4.0.0'
39	);
40
41	return PostContentUtils::getBody($post);
42	}
43
44	/**
45	* Get the body for the audio, ready to be sent to the BeyondWords API.
46	*
47	* The following rules are applied:
48	*
49	* Main body content entered in WordPress
50	* + Optionally filtered using [SpeechKit-Start]/[SpeechKit-Stop] "shortcodes"
51	* + With registered content filters FROM OTHER PLUGINS applied
52	* + Optionally prepended with the Post excerpt
53	* + Optionally filtered using the beyondwords_content filter
54	*
55	* @SuppressWarnings(PHPMD.LongVariable)
56	*
57	* @param int\|WP_Post $post The WordPress post ID, or post object.
58	*
59	* @since 3.0.0
60	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
61	* @since 3.8.0 Exclude Gutenberg blocks with attribute { beyondwordsAudio: false }
62	* @since 4.0.0 Renamed from PostContentUtils::getSourceTextForAudio() to PostContentUtils::getBody()
63	*
64	* @return string The body (the processed $post->post_content).
65	*/
66	public static function getBody($post)
67	{
68	global $beyondwords_wordpress_plugin;
69
70	$post = get_post($post);
71
72	if (!($post instanceof \WP_Post)) {
73	throw new \Exception('Post Not Found');
74	}
75
76	$content = PostContentUtils::getContentWithoutExcludedBlocks($post);
77
78	// If SpeechKit-Start/Stop tags are present then use the content within them
79	// @deprecated v3.0.0: publishers should use the beyondwords_content filter instead.
80	$regex = '/\[SpeechKit-Start\](.*?)\[SpeechKit-Stop\]/s';
81
82	if (preg_match_all($regex, $content, $match, PREG_PATTERN_ORDER) > 0) {
83	$content = implode(' ', $match[1]);
84	}
85
86	// Temporarily remove our Player filter, to exclude the player <div>
87	if ($beyondwords_wordpress_plugin && isset($beyondwords_wordpress_plugin->player)) {
88	remove_filter('the_content', array($beyondwords_wordpress_plugin->player, 'addPlayerToContent'));
89	}
90
91	// Apply other standard WordPress filters to handle shortcodes etc
92	$content = apply_filters('the_content', $content);
93
94	// Add our Player filter back in again
95	if ($beyondwords_wordpress_plugin && isset($beyondwords_wordpress_plugin->player)) {
96	add_filter('the_content', array($beyondwords_wordpress_plugin->player, 'addPlayerToContent'));
97	}
98
99	// TODO maybe we need to do something like this to deal with Shortcodes a little better?
100	// $content = wp_strip_all_tags(apply_filters('the_content', $content));
101
102	// Trim to remove trailing newlines – common for WordPress content
103	$content = trim($content);
104
105	/**
106	* Filters the content body we send for audio processing.
107	*
108	* @since 4.0.0
109	*
110	* @param string $content The post content.
111	* @param int $postId The post ID.
112	*/
113	$content = apply_filters('beyondwords_content', $content, $post->ID);
114
115	return $content;
116	}
117
118	/**
119	* Get the summary for the audio content, ready to be sent to the BeyondWords API.
120	*
121	* @param int\|WP_Post $post The WordPress post ID, or post object.
122	*
123	* @since 4.0.0
124	*
125	* @return string The summary.
126	*/
127	public static function getSummary($post)
128	{
129	$post = get_post($post);
130
131	if (!($post instanceof \WP_Post)) {
132	throw new \Exception('Post Not Found');
133	}
134
135	$summary = null;
136
137	// Optionally send the excerpt to the REST API, if the plugin setting has been checked
138	$prependExcerpt = get_option('beyondwords_prepend_excerpt');
139
140	if ($prependExcerpt && has_excerpt($post)) {
141	// Escape characters
142	$summary = htmlentities($post->post_excerpt, ENT_QUOTES \| ENT_XHTML);
143	// Apply WordPress filters
144	$summary = apply_filters('get_the_excerpt', $summary);
145	// Convert line breaks into paragraphs
146	$summary = trim(wpautop($summary));
147	}
148
149	return $summary;
150	}
151
152	/**
153	* Get the segments for the audio content, ready to be sent to the BeyondWords API.
154	*
155	* THIS METHOD IS CURRENTLY NOT IN USE. Segments cannot currently include HTML
156	* formatting tags such as <strong> and <em> so we do not pass segments, we pass
157	* a HTML string as the body param instead.
158	*
159	* @param int\|WP_Post $post The WordPress post ID, or post object.
160	*
161	* @since 4.0.0
162	*
163	* @return array\|null The segments.
164	*/
165	public static function getSegments($post)
166	{
167	if (! has_blocks($post)) {
168	return null;
169	}
170
171	$titleSegment = (object) [
172	'section' => 'title',
173	'text' => get_the_title($post),
174	];
175
176	$summarySegment = (object) [
177	'section' => 'summary',
178	'text' => PostContentUtils::getSummary($post),
179	];
180
181	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
182
183	$bodySegments = array_map(function ($block) {
184	$marker = null;
185
186	if (isset($block['attrs']) && isset($block['attrs']['beyondwordsMarker'])) {
187	$marker = $block['attrs']['beyondwordsMarker'];
188	}
189
190	return (object) [
191	'section' => 'body',
192	'marker' => $marker,
193	'text' => trim(render_block($block)),
194	];
195	}, $blocks);
196
197	// Merge title, summary and body segments
198	$segments = array_values(array_merge([$titleSegment], [$summarySegment], $bodySegments));
199
200	// TODO Consider removing this when API can handle it
201	// Remove any segments with empty text
202	$segments = array_values(array_filter($segments, function ($segment) {
203	return (! empty($segment->text));
204	}));
205
206	return $segments;
207	}
208
209	/**
210	* Get the post content without blocks which have been filtered.
211	*
212	* We have added buttons into the Gutenberg editor to optionally exclude selected
213	* blocks from the source text for audio.
214	*
215	* This method filters all blocks, removing any which have been excluded.
216	*
217	* @param int\|WP_Post $post The WordPress post ID, or post object.
218	*
219	* @since 3.8.0
220	* @since 4.0.0 Replace for loop with array_reduce
221	*
222	* @return string The post body without excluded blocks.
223	*/
224	public static function getContentWithoutExcludedBlocks($post)
225	{
226	if (! has_blocks($post)) {
227	return trim($post->post_content);
228	}
229
230	$blocks = parse_blocks($post->post_content);
231
232	if (! count($blocks)) {
233	return trim($post->post_content);
234	}
235
236	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
237
238	return array_reduce($blocks, static function ($carry, $block) {
239	return $carry . trim(render_block($block));
240	}, '');
241	}
242
243	/**
244	* Get audio-enabled blocks.
245	*
246	* @param int\|WP_Post $post The WordPress post ID, or post object.
247	*
248	* @since 4.0.0
249	*
250	* @return array The blocks.
251	*/
252	public static function getAudioEnabledBlocks($post)
253	{
254	$post = get_post($post);
255
256	if (! ($post instanceof \WP_Post)) {
257	return [];
258	}
259
260	if (! has_blocks($post)) {
261	return [];
262	}
263
264	$allBlocks = parse_blocks($post->post_content);
265
266	$blocks = array_filter($allBlocks, function ($block) {
267	$enabled = true;
268
269	if (is_array($block['attrs']) && isset($block['attrs']['beyondwordsAudio'])) {
270	$enabled = (bool) $block['attrs']['beyondwordsAudio'];
271	}
272
273	return $enabled;
274	});
275
276	/**
277	* Filters the audio-enabled blocks for a post.
278	*
279	* @since 4.0.0
280	*
281	* @param array $blocks The audio-enabled post blocks.
282	* @param array $allBlocks All post blocks including those with audio disabled.
283	* @param int $postId The post ID.
284	*/
285	$blocks = apply_filters('beyondwords_post_audio_enabled_blocks', $blocks, $allBlocks, $post->ID);
286
287	return $blocks;
288	}
289
290	/**
291	* Get the body param we pass to the API.
292	*
293	* @since 3.0.0
294	* @since 3.3.0 Added metadata to aid custom playlist generation.
295	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils.
296	* @since 3.10.4 Rename `published_at` API param to `publish_date`.
297	* @since 4.0.0 Use new API params.
298	* @since 4.0.3 Ensure `image_url` is always a string.
299	*
300	* @static
301	* @param int $postId WordPress Post ID.
302	*
303	* @return Response
304	**/
305	public static function getBodyJson($postId)
306	{
307	// https://beyondwords.postman.co/workspace/BeyondWords-Developers~169d849b-f6e1-4a78-8b66-5c87969588fc/request/23751538-99727151-c025-43b3-bd12-66951a042e2c
308	$body = [
309	'type' => 'auto_segment',
310	'title' => get_the_title($postId),
311	'summary' => PostContentUtils::getSummary($postId),
312	'body' => PostContentUtils::getBody($postId),
313	'source_url' => get_the_permalink($postId),
314	'source_id' => strval($postId),
315	'author' => PostContentUtils::getAuthorName($postId),
316	'image_url' => strval(get_the_post_thumbnail_url($postId, [240, 240])),
317	'metadata' => PostContentUtils::getMetadata($postId),
318	'published' => true,
319	'publish_date' => get_post_time(PostContentUtils::DATE_FORMAT, true, $postId),
320	];
321
322	$status = get_post_status($postId);
323
324	/*
325	* If the post status is "pending" then we send { published: false } to
326	* the BeyondWords API, to prevent the generated audio from being
327	* published in playlists.
328	*
329	* We also omit { publish_date } because get_post_time() returns `false`
330	* for posts which are "Pending Review".
331	*/
332	if ($status === 'pending') {
333	$body['published'] = false;
334	unset($body['publish_date']);
335	}
336
337	$bodyVoiceId = intval(get_post_meta($postId, 'beyondwords_body_voice_id', true));
338
339	if ($bodyVoiceId > 0) {
340	$body['body_voice_id'] = $bodyVoiceId;
341	}
342
343	$titleVoiceId = intval(get_post_meta($postId, 'beyondwords_title_voice_id', true));
344
345	if ($titleVoiceId > 0) {
346	$body['title_voice_id'] = $titleVoiceId;
347	}
348
349	$summaryVoiceId = intval(get_post_meta($postId, 'beyondwords_summary_voice_id', true));
350
351	if ($summaryVoiceId > 0) {
352	$body['summary_voice_id'] = $summaryVoiceId;
353	}
354
355	/**
356	* Filters the body params we send to the BeyondWords API when
357	* processing audio.
358	*
359	* @since 4.0.0
360	*
361	* @param array $body The params we send to the BeyondWords API.
362	* @param array $postId WordPress post ID.
363	*/
364	$body = apply_filters('beyondwords_body_params', $body, $postId);
365
366	return wp_json_encode($body);
367	}
368
369	/**
370	* Get the post metadata to send with BeyondWords API requests.
371	*
372	* The metadata key is defined by the BeyondWords API as "A custom object
373	* for storing meta information".
374	*
375	* The metadata values are used to create filters for playlists in the
376	* BeyondWords dashboard.
377	*
378	* We currently only include taxonomies by default, and the output of this
379	* method can be filtered using the `beyondwords_post_metadata` filter.
380	*
381	* @since 3.3.0
382	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
383	*
384	* @param int $postId Post ID.
385	*
386	* @return array
387	*/
388	public static function getMetadata($postId)
389	{
390	$metadata = new \stdClass();
391
392	$taxonomy = PostContentUtils::getAllTaxonomiesAndTerms($postId);
393
394	if (count((array)$taxonomy)) {
395	$metadata->taxonomy = $taxonomy;
396	}
397
398	/**
399	* Filters the post metadata sent to the BeyondWords API.
400	*
401	* @since 3.3.0
402	*
403	* @param object $metadata Post metadata. Defaults to the taxonomies and terms assigned to the post.
404	* @param int $postId Post ID.
405	*/
406	$metadata = apply_filters('beyondwords_post_metadata', $metadata, $postId);
407
408	return $metadata;
409	}
410
411	/**
412	* Get all taxonomies, and their selected terms, for a post.
413	*
414	* Returns an associative array of taxonomy names and terms.
415	*
416	* For example:
417	*
418	* array(
419	* "categories" => array("Category 1"),
420	* "post_tag" => array("Tag 1", "Tag 2", "Tag 3"),
421	* )
422	*
423	* @since 3.3.0
424	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
425	*
426	* @param int $postId Post ID.
427	*
428	* @return array
429	*/
430	public static function getAllTaxonomiesAndTerms($postId)
431	{
432	$postType = get_post_type($postId);
433
434	$postTypeTaxonomies = get_object_taxonomies($postType);
435
436	$taxonomies = new \stdClass();
437
438	foreach ($postTypeTaxonomies as $postTypeTaxonomy) {
439	$terms = get_the_terms($postId, $postTypeTaxonomy);
440
441	if (! empty($terms) && ! is_wp_error($terms)) {
442	$taxonomies->{(string)$postTypeTaxonomy} = wp_list_pluck($terms, 'name');
443	}
444	}
445
446	return $taxonomies;
447	}
448
449	/**
450	* Get author name for a post.
451	*
452	* @since 3.10.4
453	*
454	* @param int $postId Post ID.
455	*
456	* @return string
457	*/
458	public static function getAuthorName($postId)
459	{
460	$authorId = get_post_field('post_author', $postId);
461
462	return get_the_author_meta('display_name', $authorId);
463	}
464	}
465