speechkit / src / Component / Post / PostContentUtils.php

PostContentUtils.php in BeyondWords – AI audio for publishers 6.0.3, at src/Component/Post/PostContentUtils.php

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