speechkit / src / Component / Post / PostContentUtils.php

PostContentUtils.php in BeyondWords – AI audio for publishers 4.6.1, at src/Component/Post/PostContentUtils.php

630 lines 20.2 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 <[email protected]>
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 $post The WordPress post ID, or post object.
27	*
28	* @since 4.6.0
29	*
30	* @return string The content body param.
31	*/
32	public static function getContentBody($post)
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	* The following rules are applied:
56	*
57	* Main post body content entered in WordPress
58	* + Optionally filtered using [SpeechKit-Start]/[SpeechKit-Stop] "shortcodes"
59	* + With registered content filters FROM OTHER PLUGINS applied
60	* + Optionally prepended with the Post excerpt
61	* + Optionally filtered using the beyondwords_content filter
62	*
63	* @SuppressWarnings(PHPMD.LongVariable)
64	*
65	* @param int\|WP_Post $post The WordPress post ID, or post object.
66	*
67	* @since 3.0.0
68	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
69	* @since 3.8.0 Exclude Gutenberg blocks with attribute { beyondwordsAudio: false }
70	* @since 4.0.0 Renamed from PostContentUtils::getSourceTextForAudio() to PostContentUtils::getBody()
71	* @since 4.6.0 Renamed from PostContentUtils::getBody() to PostContentUtils::getPostBody()
72	*
73	* @return string The body (the processed $post->post_content).
74	*/
75	public static function getPostBody($post)
76	{
77	$post = get_post($post);
78
79	if (!($post instanceof \WP_Post)) {
80	throw new \Exception(esc_html__('Post Not Found', 'speechkit'));
81	}
82
83	$content = PostContentUtils::getContentWithoutExcludedBlocks($post);
84
85	// If SpeechKit-Start/Stop tags are present then use the content within them
86	// @deprecated v3.0.0: publishers should use the beyondwords_content filter instead.
87	$regex = '/\[SpeechKit-Start\](.*?)\[SpeechKit-Stop\]/s';
88
89	if (preg_match_all($regex, $content, $match, PREG_PATTERN_ORDER) > 0) {
90	$content = implode(' ', $match[1]);
91	}
92
93	// Apply other standard WordPress filters to handle shortcodes etc
94	$content = apply_filters('the_content', $content);
95
96	// Trim to remove trailing newlines – common for WordPress content
97	$content = trim($content);
98
99	/**
100	* Filters the content body we send for audio processing.
101	*
102	* Scheduled for removal in plugin version 5.0.0.
103	*
104	* @since 4.0.0
105	*
106	* @deprecated 4.3.0 Set the 'body' key in beyondwords_content_params instead.
107	*
108	* @param string $content The post content.
109	* @param int $postId The post ID.
110	*/
111	$content = apply_filters('beyondwords_content', $content, $post->ID);
112
113	return $content;
114	}
115
116	/**
117	* Get the post summary wrapper format.
118	*
119	* This is a <div> with optional attributes depending on the BeyondWords
120	* data of the post.
121	*
122	* @param int\|WP_Post $post The WordPress post ID, or post object.
123	*
124	* @since 4.6.0
125	*
126	* @return string The summary wrapper <div>.
127	*/
128	public static function getPostSummaryWrapperFormat($post)
129	{
130	$post = get_post($post);
131
132	if (!($post instanceof \WP_Post)) {
133	throw new \Exception(esc_html__('Post Not Found', 'speechkit'));
134	}
135
136	$summaryVoiceId = intval(get_post_meta($post->ID, 'beyondwords_summary_voice_id', true));
137
138	if ($summaryVoiceId > 0) {
139	return '<div data-beyondwords-summary="true" data-beyondwords-voice-id="' . $summaryVoiceId . '">%s</div>';
140	}
141
142	return '<div data-beyondwords-summary="true">%s</div>';
143	}
144
145	/**
146	* Get the post summary for the audio content.
147	*
148	* @param int\|WP_Post $post The WordPress post ID, or post object.
149	*
150	* @since 4.0.0
151	* @since 4.6.0 Renamed from PostContentUtils::getSummary() to PostContentUtils::getPostSummary()
152	*
153	* @return string The summary.
154	*/
155	public static function getPostSummary($post)
156	{
157	$post = get_post($post);
158
159	if (!($post instanceof \WP_Post)) {
160	throw new \Exception(esc_html__('Post Not Found', 'speechkit'));
161	}
162
163	$summary = null;
164
165	// Optionally send the excerpt to the REST API, if the plugin setting has been checked
166	$prependExcerpt = get_option('beyondwords_prepend_excerpt');
167
168	if ($prependExcerpt && has_excerpt($post)) {
169	// Escape characters
170	$summary = htmlentities($post->post_excerpt, ENT_QUOTES \| ENT_XHTML);
171	// Apply WordPress filters
172	$summary = apply_filters('get_the_excerpt', $summary);
173	// Convert line breaks into paragraphs
174	$summary = trim(wpautop($summary));
175	}
176
177	return $summary;
178	}
179
180	/**
181	* Get the segments for the audio content, ready to be sent to the BeyondWords API.
182	*
183	* @codeCoverageIgnore
184	* THIS METHOD IS CURRENTLY NOT IN USE. Segments cannot currently include HTML
185	* formatting tags such as <strong> and <em> so we do not pass segments, we pass
186	* a HTML string as the body param instead.
187	*
188	* @param int\|WP_Post $post The WordPress post ID, or post object.
189	*
190	* @since 4.0.0
191	*
192	* @return array\|null The segments.
193	*/
194	public static function getSegments($post)
195	{
196	if (! has_blocks($post)) {
197	return null;
198	}
199
200	$titleSegment = (object) [
201	'section' => 'title',
202	'text' => get_the_title($post),
203	];
204
205	$summarySegment = (object) [
206	'section' => 'summary',
207	'text' => PostContentUtils::getPostSummary($post),
208	];
209
210	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
211
212	$bodySegments = array_map(function ($block) {
213	$marker = null;
214
215	if (isset($block['attrs']) && isset($block['attrs']['beyondwordsMarker'])) {
216	$marker = $block['attrs']['beyondwordsMarker'];
217	}
218
219	return (object) [
220	'section' => 'body',
221	'marker' => $marker,
222	'text' => trim(render_block($block)),
223	];
224	}, $blocks);
225
226	// Merge title, summary and body segments
227	$segments = array_values(array_merge([$titleSegment], [$summarySegment], $bodySegments));
228
229	// Remove any segments with empty text
230	$segments = array_values(array_filter($segments, function ($segment) {
231	return (! empty($segment->text));
232	}));
233
234	return $segments;
235	}
236
237	/**
238	* Get the post content without blocks which have been filtered.
239	*
240	* We have added buttons into the Gutenberg editor to optionally exclude selected
241	* blocks from the source text for audio.
242	*
243	* This method filters all blocks, removing any which have been excluded.
244	*
245	* @param int\|WP_Post $post The WordPress post ID, or post object.
246	*
247	* @since 3.8.0
248	* @since 4.0.0 Replace for loop with array_reduce
249	*
250	* @return string The post body without excluded blocks.
251	*/
252	public static function getContentWithoutExcludedBlocks($post)
253	{
254	if (! has_blocks($post)) {
255	return trim($post->post_content);
256	}
257
258	$blocks = parse_blocks($post->post_content);
259	$output = '';
260
261	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
262
263	foreach ($blocks as $block) {
264	$marker = $block['attrs']['beyondwordsMarker'] ?? '';
265
266	$output .= PostContentUtils::addMarkerAttribute(
267	render_block($block),
268	$marker
269	);
270	}
271
272	return $output;
273	}
274
275	/**
276	* Get audio-enabled blocks.
277	*
278	* @param int\|WP_Post $post The WordPress post ID, or post object.
279	*
280	* @since 4.0.0
281	*
282	* @return array The blocks.
283	*/
284	public static function getAudioEnabledBlocks($post)
285	{
286	$post = get_post($post);
287
288	if (! ($post instanceof \WP_Post)) {
289	return [];
290	}
291
292	if (! has_blocks($post)) {
293	return [];
294	}
295
296	$allBlocks = parse_blocks($post->post_content);
297
298	$blocks = array_filter($allBlocks, function ($block) {
299	$enabled = true;
300
301	if (is_array($block['attrs']) && isset($block['attrs']['beyondwordsAudio'])) {
302	$enabled = (bool) $block['attrs']['beyondwordsAudio'];
303	}
304
305	return $enabled;
306	});
307
308	/**
309	* Filters the audio-enabled blocks for a post.
310	*
311	* Scheduled for removal in plugin version 5.0.0.
312	*
313	* @since 4.0.0
314	*
315	* @deprecated 4.3.0 Replace with {@link https://docs.beyondwords.io/docs-and-guides/content/filter-content}.
316	*
317	* @param array $blocks The audio-enabled post blocks.
318	* @param array $allBlocks All post blocks including those with audio disabled.
319	* @param int $postId The post ID.
320	*/
321	$blocks = apply_filters('beyondwords_post_audio_enabled_blocks', $blocks, $allBlocks, $post->ID);
322
323	return $blocks;
324	}
325
326	/**
327	* Get the body param we pass to the API.
328	*
329	* @since 3.0.0 Introduced as getBodyJson.
330	* @since 3.3.0 Added metadata to aid custom playlist generation.
331	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils.
332	* @since 3.10.4 Rename `published_at` API param to `publish_date`.
333	* @since 4.0.0 Use new API params.
334	* @since 4.0.3 Ensure `image_url` is always a string.
335	* @since 4.3.0 Rename from getBodyJson to getContentParams.
336	* @since 4.6.0 Remove summary param & prepend body with summary.
337	*
338	* @static
339	* @param int $postId WordPress Post ID.
340	*
341	* @return Response
342	**/
343	public static function getContentParams($postId)
344	{
345	$body = [
346	'type' => 'auto_segment',
347	'title' => get_the_title($postId),
348	'body' => PostContentUtils::getContentBody($postId),
349	'source_url' => get_the_permalink($postId),
350	'source_id' => strval($postId),
351	'author' => PostContentUtils::getAuthorName($postId),
352	'image_url' => strval(wp_get_original_image_url(get_post_thumbnail_id($postId))),
353	'metadata' => PostContentUtils::getMetadata($postId),
354	'published' => true,
355	'publish_date' => get_post_time(PostContentUtils::DATE_FORMAT, true, $postId),
356	];
357
358	$status = get_post_status($postId);
359
360	/*
361	* If the post status is "pending" then we send { published: false } to
362	* the BeyondWords API, to prevent the generated audio from being
363	* published in playlists.
364	*
365	* We also omit { publish_date } because get_post_time() returns `false`
366	* for posts which are "Pending Review".
367	*/
368	if ($status === 'pending') {
369	$body['published'] = false;
370	unset($body['publish_date']);
371	}
372
373	$bodyVoiceId = intval(get_post_meta($postId, 'beyondwords_body_voice_id', true));
374
375	if ($bodyVoiceId > 0) {
376	$body['body_voice_id'] = $bodyVoiceId;
377	}
378
379	$titleVoiceId = intval(get_post_meta($postId, 'beyondwords_title_voice_id', true));
380
381	if ($titleVoiceId > 0) {
382	$body['title_voice_id'] = $titleVoiceId;
383	}
384
385	/**
386	* Filters the params we send to the BeyondWords API 'content' endpoint.
387	*
388	* Scheduled for removal in plugin version 5.0.0.
389	*
390	* @since 4.0.0
391	*
392	* @deprecated 4.3.0 Replaced with beyondwords_content_params.
393	*
394	* @param array $body The params we send to the BeyondWords API.
395	* @param array $postId WordPress post ID.
396	*/
397	$body = apply_filters('beyondwords_body_params', $body, $postId);
398
399	/**
400	* Filters the params we send to the BeyondWords API 'content' endpoint.
401	*
402	* @since 4.0.0 Introduced as beyondwords_body_params
403	* @since 4.3.0 Renamed from beyondwords_body_params to beyondwords_content_params
404	*
405	* @param array $body The params we send to the BeyondWords API.
406	* @param array $postId WordPress post ID.
407	*/
408	$body = apply_filters('beyondwords_content_params', $body, $postId);
409
410	return wp_json_encode($body);
411	}
412
413	/**
414	* Get the post metadata to send with BeyondWords API requests.
415	*
416	* The metadata key is defined by the BeyondWords API as "A custom object
417	* for storing meta information".
418	*
419	* The metadata values are used to create filters for playlists in the
420	* BeyondWords dashboard.
421	*
422	* We currently only include taxonomies by default, and the output of this
423	* method can be filtered using the `beyondwords_post_metadata` filter.
424	*
425	* @since 3.3.0
426	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
427	*
428	* @param int $postId Post ID.
429	*
430	* @return array
431	*/
432	public static function getMetadata($postId)
433	{
434	$metadata = new \stdClass();
435
436	$taxonomy = PostContentUtils::getAllTaxonomiesAndTerms($postId);
437
438	if (count((array)$taxonomy)) {
439	$metadata->taxonomy = $taxonomy;
440	}
441
442	/**
443	* Filters the post metadata sent to the BeyondWords API.
444	*
445	* Scheduled for removal in plugin version 5.0.0.
446	*
447	* @since 3.3.0
448	*
449	* @deprecated 4.3.0 Set the 'metadata' key in beyondwords_content_params instead.
450	*
451	* @param object $metadata Post metadata. Defaults to the taxonomies and terms assigned to the post.
452	* @param int $postId Post ID.
453	*/
454	$metadata = apply_filters('beyondwords_post_metadata', $metadata, $postId);
455
456	return $metadata;
457	}
458
459	/**
460	* Get all taxonomies, and their selected terms, for a post.
461	*
462	* Returns an associative array of taxonomy names and terms.
463	*
464	* For example:
465	*
466	* array(
467	* "categories" => array("Category 1"),
468	* "post_tag" => array("Tag 1", "Tag 2", "Tag 3"),
469	* )
470	*
471	* @since 3.3.0
472	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
473	*
474	* @param int $postId Post ID.
475	*
476	* @return array
477	*/
478	public static function getAllTaxonomiesAndTerms($postId)
479	{
480	$postType = get_post_type($postId);
481
482	$postTypeTaxonomies = get_object_taxonomies($postType);
483
484	$taxonomies = new \stdClass();
485
486	foreach ($postTypeTaxonomies as $postTypeTaxonomy) {
487	$terms = get_the_terms($postId, $postTypeTaxonomy);
488
489	if (! empty($terms) && ! is_wp_error($terms)) {
490	$taxonomies->{(string)$postTypeTaxonomy} = wp_list_pluck($terms, 'name');
491	}
492	}
493
494	return $taxonomies;
495	}
496
497	/**
498	* Get author name for a post.
499	*
500	* @since 3.10.4
501	*
502	* @param int $postId Post ID.
503	*
504	* @return string
505	*/
506	public static function getAuthorName($postId)
507	{
508	$authorId = get_post_field('post_author', $postId);
509
510	return get_the_author_meta('display_name', $authorId);
511	}
512
513	/**
514	* Add data-beyondwords-marker attribute to the root elements in a HTML
515	* string (typically the rendered HTML of a single block).
516	*
517	* Checks to see whether we can use WP_HTML_Tag_Processor, or whether we
518	* fall back to using DOMDocument to add the marker.
519	*
520	* @since 4.2.2
521	*
522	* @param string $html HTML.
523	* @param string $marker Marker UUID.
524	*
525	* @return string HTML.
526	*/
527	public static function addMarkerAttribute($html, $marker)
528	{
529	if (! $marker) {
530	return $html;
531	}
532
533	// Prefer WP_HTML_Tag_Processor, introduced in WordPress 6.2
534	if (class_exists('WP_HTML_Tag_Processor')) {
535	return PostContentUtils::addMarkerAttributeWithHTMLTagProcessor($html, $marker);
536	} else {
537	return PostContentUtils::addMarkerAttributeWithDOMDocument($html, $marker);
538	}
539	}
540
541	/**
542	* Add data-beyondwords-marker attribute to the root elements in a HTML
543	* string using WP_HTML_Tag_Processor.
544	*
545	* @since 4.0.0
546	* @since 4.2.2 Moved from src/Component/Post/BlockAttributes/BlockAttributes.php
547	* to src/Component/Post/PostContentUtils.php
548	*
549	* @param string $html HTML.
550	* @param string $marker Marker UUID.
551	*
552	* @return string HTML.
553	*/
554	public static function addMarkerAttributeWithHTMLTagProcessor($html, $marker)
555	{
556	// https://github.com/WordPress/gutenberg/pull/42485
557	$tags = new \WP_HTML_Tag_Processor($html);
558
559	if ($tags->next_tag()) {
560	$tags->set_attribute('data-beyondwords-marker', $marker);
561	}
562
563	return strval($tags);
564	}
565
566	/**
567	* Add data-beyondwords-marker attribute to the root elements in a HTML
568	* string using DOMDocument.
569	*
570	* This is a fallback, since WP_HTML_Tag_Processor was only shipped with
571	* WordPress 6.2 on 19 April 2023.
572	*
573	* https://make.wordpress.org/core/2022/10/13/whats-new-in-gutenberg-14-3-12-october/
574	*
575	* Note: It is not ideal to do all the $bodyElement/$fullHtml processing
576	* in this method, but without it DOMDocument does not work as expected if
577	* there is more than 1 root element. The approach here has been taken from
578	* some historic Gutenberg code before they implemented WP_HTML_Tag_Processor:
579	*
580	* https://github.com/WordPress/gutenberg/blob/6671cef1179412a2bbd4969cbbc82705c7f69bac/lib/block-supports/index.php
581	*
582	* @since 4.0.0
583	* @since 4.2.2 Moved from src/Component/Post/BlockAttributes/BlockAttributes.php
584	* to src/Component/Post/PostContentUtils.php
585	*
586	* @param string $html HTML.
587	* @param string $marker Marker UUID.
588	*
589	* @return string HTML.
590	*/
591	public static function addMarkerAttributeWithDOMDocument($html, $marker)
592	{
593	$dom = new \DOMDocument('1.0', 'utf-8');
594
595	$wrappedHtml =
596	'<html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></head><body>'
597	. $html
598	. '</body></html>';
599
600	$success = $dom->loadHTML($wrappedHtml, LIBXML_HTML_NODEFDTD \| LIBXML_COMPACT);
601
602	if (! $success) {
603	return $html;
604	}
605
606	// Structure is like `<html><head/><body/></html>`, so body is the `lastChild` of our document.
607	$bodyElement = $dom->documentElement->lastChild;
608
609	$xpath = new \DOMXPath($dom);
610	$blockRoot = $xpath->query('./*', $bodyElement)[0];
611
612	if (empty($blockRoot)) {
613	return $html;
614	}
615
616	$blockRoot->setAttribute('data-beyondwords-marker', $marker);
617
618	// Avoid using `$dom->saveHtml( $node )` because the node results may not produce consistent
619	// whitespace. Saving the root HTML `$dom->saveHtml()` prevents this behavior.
620	$fullHtml = $dom->saveHtml();
621
622	// Find the <body> open/close tags. The open tag needs to be adjusted so we get inside the tag
623	// and not the tag itself.
624	$start = strpos($fullHtml, '<body>', 0) + strlen('<body>');
625	$end = strpos($fullHtml, '</body>', $start);
626
627	return trim(substr($fullHtml, $start, $end - $start));
628	}
629	}
630