speechkit / src / Component / Post / PostContentUtils.php

PostContentUtils.php in BeyondWords – AI audio for publishers 6.3.0, at src/Component/Post/PostContentUtils.php

403 lines 13.0 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	defined('ABSPATH') \|\| exit;
16
17	class PostContentUtils
18	{
19	public const DATE_FORMAT = 'Y-m-d\TH:i:s\Z';
20
21	/**
22	* Get the content "body" param for the audio, ready to be sent to the
23	* BeyondWords API.
24	*
25	* From API version 1.1 the "summary" param is going to be used differently,
26	* so for WordPress we now prepend the WordPress excerpt to the "body" param.
27	*
28	* @param int\|\WP_Post $post The WordPress post ID, or post object.
29	*
30	* @since 4.6.0
31	*
32	* @return string The content body param.
33	*/
34	public static function getContentBody(int\|\WP_Post $post): string\|null
35	{
36	$post = get_post($post);
37
38	if (!($post instanceof \WP_Post)) {
39	throw new \Exception(esc_html__('Post Not Found', 'speechkit'));
40	}
41
42	$summary = PostContentUtils::getPostSummary($post);
43	$body = PostContentUtils::getPostBody($post);
44
45	if ($summary) {
46	$format = PostContentUtils::getPostSummaryWrapperFormat($post);
47
48	$body = sprintf($format, $summary) . $body;
49	}
50
51	return $body;
52	}
53
54	/**
55	* Get the post body for the audio content.
56	*
57	* @since 3.0.0
58	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
59	* @since 3.8.0 Exclude Gutenberg blocks with attribute { beyondwordsAudio: false }
60	* @since 4.0.0 Renamed from PostContentUtils::getSourceTextForAudio() to PostContentUtils::getBody()
61	* @since 4.6.0 Renamed from PostContentUtils::getBody() to PostContentUtils::getPostBody()
62	* @since 4.7.0 Remove wpautop filter for block editor API requests.
63	* @since 5.0.0 Remove SpeechKit-Start shortcode.
64	* @since 5.0.0 Remove beyondwords_content filter.
65	*
66	* @param int\|\WP_Post $post The WordPress post ID, or post object.
67	*
68	* @return string The body (the processed $post->post_content).
69	*/
70	public static function getPostBody(int\|\WP_Post $post): string\|null
71	{
72	$post = get_post($post);
73
74	if (!($post instanceof \WP_Post)) {
75	throw new \Exception(esc_html__('Post Not Found', 'speechkit'));
76	}
77
78	$content = PostContentUtils::getContentWithoutExcludedBlocks($post);
79
80	if (has_blocks($post)) {
81	// wpautop breaks our HTML markup when block editor paragraphs are empty
82	remove_filter('the_content', 'wpautop');
83
84	// But we still want to remove empty lines
85	$content = preg_replace('/^\h*\v+/m', '', $content);
86	}
87
88	// Apply the_content filters to handle shortcodes etc
89	// phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound -- Applying core WordPress filter
90	$content = apply_filters('the_content', $content);
91
92	// Trim to remove trailing newlines – common for WordPress content
93	return trim($content);
94	}
95
96	/**
97	* Get the post summary wrapper format.
98	*
99	* This is a <div> with optional attributes depending on the BeyondWords
100	* data of the post.
101	*
102	* @param int\|\WP_Post $post The WordPress post ID, or post object.
103	*
104	* @since 4.6.0
105	*
106	* @return string The summary wrapper <div>.
107	*/
108	public static function getPostSummaryWrapperFormat(int\|\WP_Post $post): string
109	{
110	$post = get_post($post);
111
112	if (!($post instanceof \WP_Post)) {
113	throw new \Exception(esc_html__('Post Not Found', 'speechkit'));
114	}
115
116	$summaryVoiceId = intval(get_post_meta($post->ID, 'beyondwords_summary_voice_id', true));
117
118	if ($summaryVoiceId > 0) {
119	return '<div data-beyondwords-summary="true" data-beyondwords-voice-id="' . $summaryVoiceId . '">%s</div>';
120	}
121
122	return '<div data-beyondwords-summary="true">%s</div>';
123	}
124
125	/**
126	* Get the post summary for the audio content.
127	*
128	* @param int\|\WP_Post $post The WordPress post ID, or post object.
129	*
130	* @since 4.0.0
131	* @since 4.6.0 Renamed from PostContentUtils::getSummary() to PostContentUtils::getPostSummary()
132	*
133	* @return string The summary.
134	*/
135	public static function getPostSummary(int\|\WP_Post $post): string\|null
136	{
137	$post = get_post($post);
138
139	if (!($post instanceof \WP_Post)) {
140	throw new \Exception(esc_html__('Post Not Found', 'speechkit'));
141	}
142
143	$summary = null;
144
145	// Optionally send the excerpt to the REST API, if the plugin setting has been checked
146	$prependExcerpt = get_option('beyondwords_prepend_excerpt');
147
148	if ($prependExcerpt && has_excerpt($post)) {
149	// Escape characters
150	$summary = htmlentities($post->post_excerpt, ENT_QUOTES \| ENT_XHTML);
151	// Apply WordPress filters
152	// phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound -- Applying core WordPress filter
153	$summary = apply_filters('get_the_excerpt', $summary);
154	// Convert line breaks into paragraphs
155	$summary = trim(wpautop($summary));
156	}
157
158	return $summary;
159	}
160
161	/**
162	* Get the post content without blocks which have been filtered.
163	*
164	* We have added buttons into the Gutenberg editor to optionally exclude selected
165	* blocks from the source text for audio.
166	*
167	* This method filters all blocks, removing any which have been excluded.
168	*
169	* @param int\|\WP_Post $post The WordPress post ID, or post object.
170	*
171	* @since 3.8.0
172	* @since 4.0.0 Replace for loop with array_reduce
173	* @since 6.0.0 Remove beyondwordsMarker attribute from rendered blocks.
174	*
175	* @return string The post body without excluded blocks.
176	*/
177	public static function getContentWithoutExcludedBlocks(int\|\WP_Post $post): string
178	{
179	if (! has_blocks($post)) {
180	return trim($post->post_content);
181	}
182
183	$blocks = parse_blocks($post->post_content);
184	$output = '';
185
186	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
187
188	foreach ($blocks as $block) {
189	$output .= render_block($block);
190	}
191
192	return $output;
193	}
194
195	/**
196	* Get audio-enabled blocks.
197	*
198	* @param int\|\WP_Post $post The WordPress post ID, or post object.
199	*
200	* @since 4.0.0
201	* @since 5.0.0 Remove beyondwords_post_audio_enabled_blocks filter.
202	*
203	* @return array The blocks.
204	*/
205	public static function getAudioEnabledBlocks(int\|\WP_Post $post): array
206	{
207	$post = get_post($post);
208
209	if (! ($post instanceof \WP_Post)) {
210	return [];
211	}
212
213	if (! has_blocks($post)) {
214	return [];
215	}
216
217	$allBlocks = parse_blocks($post->post_content);
218
219	return array_filter($allBlocks, function ($block) {
220	$enabled = true;
221
222	if (is_array($block['attrs']) && isset($block['attrs']['beyondwordsAudio'])) {
223	$enabled = (bool) $block['attrs']['beyondwordsAudio'];
224	}
225
226	return $enabled;
227	});
228	}
229
230	/**
231	* Get the body param we pass to the API.
232	*
233	* @since 3.0.0 Introduced as getBodyJson.
234	* @since 3.3.0 Added metadata to aid custom playlist generation.
235	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils.
236	* @since 3.10.4 Rename `published_at` API param to `publish_date`.
237	* @since 4.0.0 Use new API params.
238	* @since 4.0.3 Ensure `image_url` is always a string.
239	* @since 4.3.0 Rename from getBodyJson to getContentParams.
240	* @since 4.6.0 Remove summary param & prepend body with summary.
241	* @since 5.0.0 Remove beyondwords_body_params filter.
242	* @since 6.0.0 Cast return value to string.
243	*
244	* @static
245	* @param int $postId WordPress Post ID.
246	*
247	* @return string JSON endoded params.
248	**/
249	public static function getContentParams(int $postId): array\|string
250	{
251	$body = [
252	'type' => 'auto_segment',
253	'title' => get_the_title($postId),
254	'body' => PostContentUtils::getContentBody($postId),
255	'source_url' => get_the_permalink($postId),
256	'source_id' => strval($postId),
257	'author' => PostContentUtils::getAuthorName($postId),
258	'image_url' => strval(wp_get_original_image_url(get_post_thumbnail_id($postId))),
259	'metadata' => PostContentUtils::getMetadata($postId),
260	'publish_date' => get_post_time(PostContentUtils::DATE_FORMAT, true, $postId),
261	];
262
263	$status = get_post_status($postId);
264
265	/*
266	* If the post status is draft/pending then we explicity send
267	* { published: false } to the BeyondWords API, to prevent the
268	* generated audio from being published in playlists.
269	*
270	* We also omit { publish_date } because get_post_time() returns `false`
271	* for posts which are "Pending Review".
272	*/
273	if (in_array($status, ['draft', 'pending'])) {
274	$body['published'] = false;
275	unset($body['publish_date']);
276	} elseif (get_option('beyondwords_project_auto_publish_enabled')) {
277	$body['published'] = true;
278	}
279
280	$languageCode = get_post_meta($postId, 'beyondwords_language_code', true);
281
282	if ($languageCode) {
283	$body['language'] = $languageCode;
284	}
285
286	$bodyVoiceId = intval(get_post_meta($postId, 'beyondwords_body_voice_id', true));
287
288	if ($bodyVoiceId > 0) {
289	$body['body_voice_id'] = $bodyVoiceId;
290	}
291
292	$titleVoiceId = intval(get_post_meta($postId, 'beyondwords_title_voice_id', true));
293
294	if ($titleVoiceId > 0) {
295	$body['title_voice_id'] = $titleVoiceId;
296	}
297
298	$summaryVoiceId = intval(get_post_meta($postId, 'beyondwords_summary_voice_id', true));
299
300	if ($summaryVoiceId > 0) {
301	$body['summary_voice_id'] = $summaryVoiceId;
302	}
303
304	/**
305	* Filters the params we send to the BeyondWords API 'content' endpoint.
306	*
307	* @since 4.0.0 Introduced as beyondwords_body_params
308	* @since 4.3.0 Renamed from beyondwords_body_params to beyondwords_content_params
309	*
310	* @param array $body The params we send to the BeyondWords API.
311	* @param array $postId WordPress post ID.
312	*/
313	$body = apply_filters('beyondwords_content_params', $body, $postId);
314
315	return (string) wp_json_encode($body);
316	}
317
318	/**
319	* Get the post metadata to send with BeyondWords API requests.
320	*
321	* The metadata key is defined by the BeyondWords API as "A custom object
322	* for storing meta information".
323	*
324	* The metadata values are used to create filters for playlists in the
325	* BeyondWords dashboard.
326	*
327	* We currently only include taxonomies by default, and the output of this
328	* method can be filtered using the `beyondwords_post_metadata` filter.
329	*
330	* @since 3.3.0
331	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils.
332	* @since 5.0.0 Remove beyondwords_post_metadata filter.
333	*
334	* @param int $postId Post ID.
335	*
336	* @return object The metadata object (empty if no metadata).
337	*/
338	public static function getMetadata(int $postId): array\|object
339	{
340	$metadata = new \stdClass();
341
342	$taxonomy = PostContentUtils::getAllTaxonomiesAndTerms($postId);
343
344	if (count((array)$taxonomy)) {
345	$metadata->taxonomy = $taxonomy;
346	}
347
348	return $metadata;
349	}
350
351	/**
352	* Get all taxonomies, and their selected terms, for a post.
353	*
354	* Returns an associative array of taxonomy names and terms.
355	*
356	* For example:
357	*
358	* array(
359	* "categories" => array("Category 1"),
360	* "post_tag" => array("Tag 1", "Tag 2", "Tag 3"),
361	* )
362	*
363	* @since 3.3.0
364	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
365	*
366	* @param int $postId Post ID.
367	*
368	* @return object The taxonomies object (empty if no taxonomies).
369	*/
370	public static function getAllTaxonomiesAndTerms(int $postId): array\|object
371	{
372	$postType = get_post_type($postId);
373
374	$postTypeTaxonomies = get_object_taxonomies($postType);
375
376	$taxonomies = new \stdClass();
377
378	foreach ($postTypeTaxonomies as $postTypeTaxonomy) {
379	$terms = get_the_terms($postId, $postTypeTaxonomy);
380
381	if (! empty($terms) && ! is_wp_error($terms)) {
382	$taxonomies->{(string)$postTypeTaxonomy} = wp_list_pluck($terms, 'name');
383	}
384	}
385
386	return $taxonomies;
387	}
388
389	/**
390	* Get author name for a post.
391	*
392	* @since 3.10.4
393	*
394	* @param int $postId Post ID.
395	*/
396	public static function getAuthorName(int $postId): string
397	{
398	$authorId = get_post_field('post_author', $postId);
399
400	return get_the_author_meta('display_name', $authorId);
401	}
402	}
403