speechkit / src / post / class-content.php

class-content.php in BeyondWords – AI audio for publishers 7.0.0, at src/post/class-content.php

433 lines 13.8 KB

Raw Download Zip

1	<?php
2
3	declare( strict_types = 1 );
4
5	namespace BeyondWords\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	* @since 7.0.0 Refactored to BeyondWords namespace with snake_case methods.
15	*/
16	defined( 'ABSPATH' ) \|\| exit;
17
18	class Content {
19
20	public const DATE_FORMAT = 'Y-m-d\TH:i:s\Z';
21
22	/**
23	* Get the content "body" param for the audio.
24	*
25	* The excerpt is prepended to the body because API v1.1 repurposed the
26	* "summary" param.
27	*
28	* @since 4.6.0
29	* @since 7.0.0 Refactored to BeyondWords namespace with snake_case methods.
30	*
31	* @param int\|\WP_Post $post The WordPress post ID, or post object.
32	*
33	* @return string The content body param.
34	*/
35	public static function get_content_body( int\|\WP_Post $post ): string\|null {
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 = self::get_post_summary( $post );
43	$body = self::get_post_body( $post );
44
45	if ( $summary ) {
46	$format = self::get_post_summary_wrapper_format( $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 Content::getSourceTextForAudio() to Content::getBody()
61	* @since 4.6.0 Renamed from Content::getBody() to Content::get_post_body()
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	* @since 7.0.0 Refactored to BeyondWords namespace with snake_case methods.
66	*
67	* @param int\|\WP_Post $post The WordPress post ID, or post object.
68	*
69	* @return string The body (the processed $post->post_content).
70	*/
71	public static function get_post_body( int\|\WP_Post $post ): string\|null {
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 = self::get_content_without_excluded_blocks( $post );
79
80	if ( has_blocks( $post ) ) {
81	// wpautop breaks our HTML markup when block editor paragraphs are empty,
82	// but we still want to remove the empty lines it would have handled.
83	remove_filter( 'the_content', 'wpautop' );
84
85	$content = preg_replace( '/^\h*\v+/m', '', $content );
86	}
87
88	// phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound -- Applying core WordPress filter
89	$content = apply_filters( 'the_content', $content );
90
91	return trim( $content );
92	}
93
94	/**
95	* Get the post summary wrapper format.
96	*
97	* @since 4.6.0
98	* @since 7.0.0 Refactored to BeyondWords namespace with snake_case methods.
99	*
100	* @param int\|\WP_Post $post The WordPress post ID, or post object.
101	*
102	* @return string The summary wrapper <div>.
103	*/
104	public static function get_post_summary_wrapper_format( int\|\WP_Post $post ): string {
105	$post = get_post( $post );
106
107	if ( ! ( $post instanceof \WP_Post ) ) {
108	throw new \Exception( esc_html__( 'Post Not Found', 'speechkit' ) );
109	}
110
111	return '<div data-beyondwords-summary="true">%s</div>';
112	}
113
114	/**
115	* Get the post summary for the audio content.
116	*
117	* @since 4.0.0
118	* @since 4.6.0 Renamed from Content::getSummary() to Content::get_post_summary()
119	* @since 7.0.0 Refactored to BeyondWords namespace with snake_case methods.
120	*
121	* @param int\|\WP_Post $post The WordPress post ID, or post object.
122	*
123	* @return string The summary.
124	*/
125	public static function get_post_summary( int\|\WP_Post $post ): string\|null {
126	$post = get_post( $post );
127
128	if ( ! ( $post instanceof \WP_Post ) ) {
129	throw new \Exception( esc_html__( 'Post Not Found', 'speechkit' ) );
130	}
131
132	$summary = null;
133
134	$prepend_excerpt = get_option( 'beyondwords_prepend_excerpt' );
135
136	if ( $prepend_excerpt && has_excerpt( $post ) ) {
137	$summary = htmlentities( $post->post_excerpt, ENT_QUOTES \| ENT_XHTML );
138	// phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound -- Applying core WordPress filter
139	$summary = apply_filters( 'get_the_excerpt', $summary );
140	$summary = trim( wpautop( $summary ) );
141	}
142
143	return $summary;
144	}
145
146	/**
147	* Get the post content without the blocks an editor excluded from audio.
148	*
149	* @since 3.8.0
150	* @since 4.0.0 Replace for loop with array_reduce
151	* @since 6.0.0 Remove beyondwordsMarker attribute from rendered blocks.
152	* @since 7.0.0 Refactored to BeyondWords namespace with snake_case methods.
153	*
154	* @param int\|\WP_Post $post The WordPress post ID, or post object.
155	*
156	* @return string The post body without excluded blocks.
157	*/
158	public static function get_content_without_excluded_blocks( int\|\WP_Post $post ): string {
159	$post = get_post( $post );
160
161	if ( ! ( $post instanceof \WP_Post ) ) {
162	throw new \Exception( esc_html__( 'Post Not Found', 'speechkit' ) );
163	}
164
165	if ( ! has_blocks( $post ) ) {
166	return trim( $post->post_content );
167	}
168
169	$output = '';
170
171	$blocks = self::get_audio_enabled_blocks( $post );
172
173	foreach ( $blocks as $block ) {
174	$output .= render_block( $block );
175	}
176
177	return $output;
178	}
179
180	/**
181	* Get audio-enabled blocks.
182	*
183	* @since 4.0.0
184	* @since 5.0.0 Remove beyondwords_post_audio_enabled_blocks filter.
185	* @since 7.0.0 Refactored to BeyondWords namespace with snake_case methods.
186	*
187	* @param int\|\WP_Post $post The WordPress post ID, or post object.
188	*
189	* @return array The blocks.
190	*/
191	public static function get_audio_enabled_blocks( int\|\WP_Post $post ): array {
192	$post = get_post( $post );
193
194	if ( ! ( $post instanceof \WP_Post ) ) {
195	return [];
196	}
197
198	if ( ! has_blocks( $post ) ) {
199	return [];
200	}
201
202	$all_blocks = parse_blocks( $post->post_content );
203
204	return array_filter(
205	$all_blocks,
206	function ( $block ) {
207	$enabled = true;
208
209	if ( is_array( $block['attrs'] ) && isset( $block['attrs']['beyondwordsAudio'] ) ) {
210	$enabled = (bool) $block['attrs']['beyondwordsAudio'];
211	}
212
213	return $enabled;
214	}
215	);
216	}
217
218	/**
219	* Get the body param we pass to the API.
220	*
221	* @since 3.0.0 Introduced as getBodyJson.
222	* @since 3.3.0 Added metadata to aid custom playlist generation.
223	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils.
224	* @since 3.10.4 Rename `published_at` API param to `publish_date`.
225	* @since 4.0.0 Use new API params.
226	* @since 4.0.3 Ensure `image_url` is always a string.
227	* @since 4.3.0 Rename from getBodyJson to getContentParams.
228	* @since 4.6.0 Remove summary param & prepend body with summary.
229	* @since 5.0.0 Remove beyondwords_body_params filter.
230	* @since 6.0.0 Cast return value to string.
231	* @since 7.0.0 Replace the `metadata` param with a flat `tags` array.
232	*
233	* @static
234	* @param int $post_id WordPress Post ID.
235	*
236	* @return string JSON encoded params.
237	**/
238	public static function get_content_params( int $post_id ): array\|string {
239	$body = [
240	'type' => 'auto_segment',
241	'title' => get_the_title( $post_id ),
242	'body' => self::get_content_body( $post_id ),
243	'source_url' => get_the_permalink( $post_id ),
244	'source_id' => strval( $post_id ),
245	'author' => self::get_author_name( $post_id ),
246	'image_url' => strval( wp_get_original_image_url( get_post_thumbnail_id( $post_id ) ) ),
247	'tags' => self::get_tags( $post_id ),
248	'publish_date' => get_post_time( self::DATE_FORMAT, true, $post_id ),
249	];
250
251	$status = get_post_status( $post_id );
252
253	// Drafts send { published: false } to keep the audio out of playlists, and
254	// omit publish_date because get_post_time() is false for pending posts.
255	if ( in_array( $status, [ 'draft', 'pending'] ) ) {
256	$body['published'] = false;
257	unset( $body['publish_date'] );
258	} else {
259	/**
260	* Filters whether generated content is auto-published to BeyondWords.
261	*
262	* Replaces the v6.x `beyondwords_project_auto_publish_enabled` setting.
263	*
264	* @since 7.0.0
265	*
266	* @param bool $auto_publish Whether to mark generated content as published.
267	* @param int $post_id WordPress post ID.
268	*/
269	$auto_publish = apply_filters( 'beyondwords_auto_publish', true, $post_id );
270
271	if ( $auto_publish ) {
272	$body['published'] = true;
273	}
274	}
275
276	// The language is never sent: a chosen voice implies it, and with no voice
277	// the project default applies. `beyondwords_language_code` is editor state only.
278	$body_voice_id = intval( get_post_meta( $post_id, 'beyondwords_body_voice_id', true ) );
279
280	if ( $body_voice_id > 0 ) {
281	$body['body_voice_id'] = $body_voice_id;
282	}
283
284	// Omitted when Source is Post (or unset) so the project default applies.
285	$source = \BeyondWords\Editor\Components\SettingsFields::get_source( $post_id );
286
287	if ( \BeyondWords\Editor\Components\SettingsFields::source_includes_script( $source ) ) {
288	$body['summarization_settings'] = [ 'enabled' => true ];
289
290	$script_template_id = intval(
291	get_post_meta( $post_id, 'beyondwords_script_template_id', true )
292	);
293
294	if ( $script_template_id > 0 ) {
295	$body['summarization_settings']['template'] = [
296	'id' => $script_template_id,
297	];
298	}
299	}
300
301	$output = get_post_meta( $post_id, 'beyondwords_output', true );
302
303	if ( in_array( $output, [ 'video', 'audio_and_video' ], true ) ) {
304	$body['video_settings'] = self::get_video_settings_params( $post_id );
305	}
306
307	/**
308	* Filters the params we send to the BeyondWords API 'content' endpoint.
309	*
310	* @since 4.0.0 Introduced as beyondwords_body_params
311	* @since 4.3.0 Renamed from beyondwords_body_params to beyondwords_content_params
312	*
313	* @param array $body The params we send to the BeyondWords API.
314	* @param array $post_id WordPress post ID.
315	*/
316	$body = apply_filters( 'beyondwords_content_params', $body, $post_id );
317
318	return (string) wp_json_encode( $body );
319	}
320
321	/**
322	* Build the `video_settings` param sent to the BeyondWords content endpoint.
323	*
324	* The backend silently skips video generation unless the payload has `enabled:
325	* true` plus non-empty `variants` and `sizes` (with dimensions), so we seed
326	* from the project defaults and layer the post's choices on top. See doc/video-settings-payload.md.
327	*
328	* @since 7.0.0
329	*
330	* @param int $post_id WordPress post ID.
331	*
332	* @return array<string, mixed> The `video_settings` param.
333	*/
334	private static function get_video_settings_params( int $post_id ): array {
335	// The post's project may be a per-post override of the global one.
336	$project_id = \BeyondWords\Post\Meta::get_project_id( $post_id );
337	$defaults = \BeyondWords\Api\Client::get_video_settings( is_numeric( $project_id ) ? (int) $project_id : null );
338	$defaults = is_array( $defaults ) ? $defaults : [];
339
340	$settings = [ 'enabled' => true ];
341
342	// The backend needs a non-empty `variants`; there is no per-post variant
343	// control, so echo the project defaults.
344	if ( ! empty( $defaults['variants'] ) && is_array( $defaults['variants'] ) ) {
345	$settings['variants'] = array_values( $defaults['variants'] );
346	}
347
348	// Echo the project sizes (the backend requires width/height), enabling
349	// only the post's chosen size when one is set.
350	$video_size = (string) get_post_meta( $post_id, 'beyondwords_video_size', true );
351	$default_sizes = ( isset( $defaults['sizes'] ) && is_array( $defaults['sizes'] ) ) ? $defaults['sizes'] : [];
352
353	$sizes = [];
354
355	foreach ( $default_sizes as $size ) {
356	if ( ! is_array( $size ) \|\| ! isset( $size['name'] ) ) {
357	continue;
358	}
359
360	$sizes[] = [
361	'name' => (string) $size['name'],
362	'width' => (int) ( $size['width'] ?? 0 ),
363	'height' => (int) ( $size['height'] ?? 0 ),
364	'enabled' => '' !== $video_size
365	? ( (string) $size['name'] === $video_size )
366	: (bool) ( $size['enabled'] ?? false ),
367	];
368	}
369
370	if ( ! empty( $sizes ) ) {
371	$settings['sizes'] = $sizes;
372	}
373
374	// Omit `template` to defer to the project default.
375	$video_template_id = intval( get_post_meta( $post_id, 'beyondwords_video_template_id', true ) );
376
377	if ( $video_template_id > 0 ) {
378	$settings['template'] = [ 'id' => $video_template_id ];
379	}
380
381	return $settings;
382	}
383
384	/**
385	* Get the taxonomy terms to send as the `tags` param.
386	*
387	* The values are used to create playlist filters in the BeyondWords dashboard.
388	*
389	* @since 3.3.0 Introduced as get_metadata(), sending a metadata.taxonomy object.
390	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils.
391	* @since 5.0.0 Remove beyondwords_post_metadata filter.
392	* @since 7.0.0 Renamed to get_tags(), returning a flat array of term names.
393	*
394	* @return string[] Term names from every taxonomy of the post type.
395	*/
396	public static function get_tags( int $post_id ): array {
397	$taxonomies = get_object_taxonomies( (string) get_post_type( $post_id ) );
398
399	$tags = [];
400
401	foreach ( $taxonomies as $taxonomy ) {
402	$terms = get_the_terms( $post_id, $taxonomy );
403
404	if ( ! empty( $terms ) && ! is_wp_error( $terms ) ) {
405	$tags = array_merge( $tags, wp_list_pluck( $terms, 'name' ) );
406	}
407	}
408
409	// Core stores term names HTML-encoded, so "R&D" would reach the API as "R&D".
410	$tags = array_map( fn( $tag ) => wp_specialchars_decode( $tag, ENT_QUOTES ), $tags );
411
412	// Terms in different taxonomies can share a name, and the API wants each tag once.
413	return array_values( array_unique( $tags ) );
414	}
415
416	/**
417	* Get author name for a post.
418	*
419	* @since 3.10.4
420	* @since 7.0.0 Refactored to BeyondWords namespace with snake_case methods.
421	*
422	* @param int $post_id Post ID.
423	*/
424	public static function get_author_name( int $post_id ): string {
425	$author_id = get_post_field( 'post_author', $post_id );
426
427	$name = get_the_author_meta( 'display_name', $author_id );
428
429	// Core stores display names HTML-encoded, so "Smith & Sons" would reach the API as "Smith & Sons".
430	return wp_specialchars_decode( $name, ENT_QUOTES );
431	}
432	}
433