speechkit / src / Component / Post / PostContentUtils.php

PostContentUtils.php in BeyondWords – AI audio for publishers 4.7.0, at src/Component/Post/PostContentUtils.php

649 lines 20.8 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 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	* @since 4.7.0 Remove wpautop filter for block editor API requests.
73	*
74	* @return string The body (the processed $post->post_content).
75	*/
76	public static function getPostBody($post)
77	{
78	$post = get_post($post);
79
80	if (!($post instanceof \WP_Post)) {
81	throw new \Exception(esc_html__('Post Not Found', 'speechkit'));
82	}
83
84	$content = PostContentUtils::getContentWithoutExcludedBlocks($post);
85
86	// If SpeechKit-Start/Stop tags are present then use the content within them
87	// @deprecated v3.0.0: publishers should use the beyondwords_content filter instead.
88	$regex = '/\[SpeechKit-Start\](.*?)\[SpeechKit-Stop\]/s';
89
90	if (preg_match_all($regex, $content, $match, PREG_PATTERN_ORDER) > 0) {
91	$content = implode(' ', $match[1]);
92	}
93
94	if (has_blocks($post)) {
95	// wpautop breaks our HTML markup when block editor paragraphs are empty
96	remove_filter('the_content', 'wpautop');
97
98	// But we still want to remove empty lines
99	$content = preg_replace('/^\h*\v+/m', '', $content);
100	}
101
102	// Apply the_content filters to handle shortcodes etc
103	$content = apply_filters('the_content', $content);
104
105	// Trim to remove trailing newlines – common for WordPress content
106	$content = trim($content);
107
108	/**
109	* Filters the content body we send for audio processing.
110	*
111	* Scheduled for removal in plugin version 5.0.0.
112	*
113	* @since 4.0.0
114	*
115	* @deprecated 4.3.0 Set the 'body' key in beyondwords_content_params instead.
116	*
117	* @param string $content The post content.
118	* @param int $postId The post ID.
119	*/
120	$content = apply_filters('beyondwords_content', $content, $post->ID);
121
122	return $content;
123	}
124
125	/**
126	* Get the post summary wrapper format.
127	*
128	* This is a <div> with optional attributes depending on the BeyondWords
129	* data of the post.
130	*
131	* @param int\|WP_Post $post The WordPress post ID, or post object.
132	*
133	* @since 4.6.0
134	*
135	* @return string The summary wrapper <div>.
136	*/
137	public static function getPostSummaryWrapperFormat($post)
138	{
139	$post = get_post($post);
140
141	if (!($post instanceof \WP_Post)) {
142	throw new \Exception(esc_html__('Post Not Found', 'speechkit'));
143	}
144
145	$summaryVoiceId = intval(get_post_meta($post->ID, 'beyondwords_summary_voice_id', true));
146
147	if ($summaryVoiceId > 0) {
148	return '<div data-beyondwords-summary="true" data-beyondwords-voice-id="' . $summaryVoiceId . '">%s</div>';
149	}
150
151	return '<div data-beyondwords-summary="true">%s</div>';
152	}
153
154	/**
155	* Get the post summary for the audio content.
156	*
157	* @param int\|WP_Post $post The WordPress post ID, or post object.
158	*
159	* @since 4.0.0
160	* @since 4.6.0 Renamed from PostContentUtils::getSummary() to PostContentUtils::getPostSummary()
161	*
162	* @return string The summary.
163	*/
164	public static function getPostSummary($post)
165	{
166	$post = get_post($post);
167
168	if (!($post instanceof \WP_Post)) {
169	throw new \Exception(esc_html__('Post Not Found', 'speechkit'));
170	}
171
172	$summary = null;
173
174	// Optionally send the excerpt to the REST API, if the plugin setting has been checked
175	$prependExcerpt = get_option('beyondwords_prepend_excerpt');
176
177	if ($prependExcerpt && has_excerpt($post)) {
178	// Escape characters
179	$summary = htmlentities($post->post_excerpt, ENT_QUOTES \| ENT_XHTML);
180	// Apply WordPress filters
181	$summary = apply_filters('get_the_excerpt', $summary);
182	// Convert line breaks into paragraphs
183	$summary = trim(wpautop($summary));
184	}
185
186	return $summary;
187	}
188
189	/**
190	* Get the segments for the audio content, ready to be sent to the BeyondWords API.
191	*
192	* @codeCoverageIgnore
193	* THIS METHOD IS CURRENTLY NOT IN USE. Segments cannot currently include HTML
194	* formatting tags such as <strong> and <em> so we do not pass segments, we pass
195	* a HTML string as the body param instead.
196	*
197	* @param int\|WP_Post $post The WordPress post ID, or post object.
198	*
199	* @since 4.0.0
200	*
201	* @return array\|null The segments.
202	*/
203	public static function getSegments($post)
204	{
205	if (! has_blocks($post)) {
206	return null;
207	}
208
209	$titleSegment = (object) [
210	'section' => 'title',
211	'text' => get_the_title($post),
212	];
213
214	$summarySegment = (object) [
215	'section' => 'summary',
216	'text' => PostContentUtils::getPostSummary($post),
217	];
218
219	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
220
221	$bodySegments = array_map(function ($block) {
222	$marker = null;
223
224	if (isset($block['attrs']) && isset($block['attrs']['beyondwordsMarker'])) {
225	$marker = $block['attrs']['beyondwordsMarker'];
226	}
227
228	return (object) [
229	'section' => 'body',
230	'marker' => $marker,
231	'text' => trim(render_block($block)),
232	];
233	}, $blocks);
234
235	// Merge title, summary and body segments
236	$segments = array_values(array_merge([$titleSegment], [$summarySegment], $bodySegments));
237
238	// Remove any segments with empty text
239	$segments = array_values(array_filter($segments, function ($segment) {
240	return (! empty($segment->text));
241	}));
242
243	return $segments;
244	}
245
246	/**
247	* Get the post content without blocks which have been filtered.
248	*
249	* We have added buttons into the Gutenberg editor to optionally exclude selected
250	* blocks from the source text for audio.
251	*
252	* This method filters all blocks, removing any which have been excluded.
253	*
254	* @param int\|WP_Post $post The WordPress post ID, or post object.
255	*
256	* @since 3.8.0
257	* @since 4.0.0 Replace for loop with array_reduce
258	*
259	* @return string The post body without excluded blocks.
260	*/
261	public static function getContentWithoutExcludedBlocks($post)
262	{
263	if (! has_blocks($post)) {
264	return trim($post->post_content);
265	}
266
267	$blocks = parse_blocks($post->post_content);
268	$output = '';
269
270	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
271
272	foreach ($blocks as $block) {
273	$marker = $block['attrs']['beyondwordsMarker'] ?? '';
274
275	$output .= PostContentUtils::addMarkerAttribute(
276	render_block($block),
277	$marker
278	);
279	}
280
281	return $output;
282	}
283
284	/**
285	* Get audio-enabled blocks.
286	*
287	* @param int\|WP_Post $post The WordPress post ID, or post object.
288	*
289	* @since 4.0.0
290	*
291	* @return array The blocks.
292	*/
293	public static function getAudioEnabledBlocks($post)
294	{
295	$post = get_post($post);
296
297	if (! ($post instanceof \WP_Post)) {
298	return [];
299	}
300
301	if (! has_blocks($post)) {
302	return [];
303	}
304
305	$allBlocks = parse_blocks($post->post_content);
306
307	$blocks = array_filter($allBlocks, function ($block) {
308	$enabled = true;
309
310	if (is_array($block['attrs']) && isset($block['attrs']['beyondwordsAudio'])) {
311	$enabled = (bool) $block['attrs']['beyondwordsAudio'];
312	}
313
314	return $enabled;
315	});
316
317	/**
318	* Filters the audio-enabled blocks for a post.
319	*
320	* Scheduled for removal in plugin version 5.0.0.
321	*
322	* @since 4.0.0
323	*
324	* @deprecated 4.3.0 Replace with {@link https://docs.beyondwords.io/docs-and-guides/content/filter-content}.
325	*
326	* @param array $blocks The audio-enabled post blocks.
327	* @param array $allBlocks All post blocks including those with audio disabled.
328	* @param int $postId The post ID.
329	*/
330	$blocks = apply_filters('beyondwords_post_audio_enabled_blocks', $blocks, $allBlocks, $post->ID);
331
332	return $blocks;
333	}
334
335	/**
336	* Get the body param we pass to the API.
337	*
338	* @since 3.0.0 Introduced as getBodyJson.
339	* @since 3.3.0 Added metadata to aid custom playlist generation.
340	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils.
341	* @since 3.10.4 Rename `published_at` API param to `publish_date`.
342	* @since 4.0.0 Use new API params.
343	* @since 4.0.3 Ensure `image_url` is always a string.
344	* @since 4.3.0 Rename from getBodyJson to getContentParams.
345	* @since 4.6.0 Remove summary param & prepend body with summary.
346	*
347	* @static
348	* @param int $postId WordPress Post ID.
349	*
350	* @return Response
351	**/
352	public static function getContentParams($postId)
353	{
354	$body = [
355	'type' => 'auto_segment',
356	'title' => get_the_title($postId),
357	'body' => PostContentUtils::getContentBody($postId),
358	'source_url' => get_the_permalink($postId),
359	'source_id' => strval($postId),
360	'author' => PostContentUtils::getAuthorName($postId),
361	'image_url' => strval(wp_get_original_image_url(get_post_thumbnail_id($postId))),
362	'metadata' => PostContentUtils::getMetadata($postId),
363	'published' => true,
364	'publish_date' => get_post_time(PostContentUtils::DATE_FORMAT, true, $postId),
365	];
366
367	$status = get_post_status($postId);
368
369	/*
370	* If the post status is "pending" then we send { published: false } to
371	* the BeyondWords API, to prevent the generated audio from being
372	* published in playlists.
373	*
374	* We also omit { publish_date } because get_post_time() returns `false`
375	* for posts which are "Pending Review".
376	*/
377	if ($status === 'pending') {
378	$body['published'] = false;
379	unset($body['publish_date']);
380	}
381
382	$bodyVoiceId = intval(get_post_meta($postId, 'beyondwords_body_voice_id', true));
383
384	if ($bodyVoiceId > 0) {
385	$body['body_voice_id'] = $bodyVoiceId;
386	}
387
388	$titleVoiceId = intval(get_post_meta($postId, 'beyondwords_title_voice_id', true));
389
390	if ($titleVoiceId > 0) {
391	$body['title_voice_id'] = $titleVoiceId;
392	}
393
394	/**
395	* Filters the params we send to the BeyondWords API 'content' endpoint.
396	*
397	* Scheduled for removal in plugin version 5.0.0.
398	*
399	* @since 4.0.0
400	*
401	* @deprecated 4.3.0 Replaced with beyondwords_content_params.
402	*
403	* @param array $body The params we send to the BeyondWords API.
404	* @param array $postId WordPress post ID.
405	*/
406	$body = apply_filters('beyondwords_body_params', $body, $postId);
407
408	/**
409	* Filters the params we send to the BeyondWords API 'content' endpoint.
410	*
411	* @since 4.0.0 Introduced as beyondwords_body_params
412	* @since 4.3.0 Renamed from beyondwords_body_params to beyondwords_content_params
413	*
414	* @param array $body The params we send to the BeyondWords API.
415	* @param array $postId WordPress post ID.
416	*/
417	$body = apply_filters('beyondwords_content_params', $body, $postId);
418
419	return wp_json_encode($body);
420	}
421
422	/**
423	* Get the post metadata to send with BeyondWords API requests.
424	*
425	* The metadata key is defined by the BeyondWords API as "A custom object
426	* for storing meta information".
427	*
428	* The metadata values are used to create filters for playlists in the
429	* BeyondWords dashboard.
430	*
431	* We currently only include taxonomies by default, and the output of this
432	* method can be filtered using the `beyondwords_post_metadata` filter.
433	*
434	* @since 3.3.0
435	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
436	*
437	* @param int $postId Post ID.
438	*
439	* @return array
440	*/
441	public static function getMetadata($postId)
442	{
443	$metadata = new \stdClass();
444
445	$taxonomy = PostContentUtils::getAllTaxonomiesAndTerms($postId);
446
447	if (count((array)$taxonomy)) {
448	$metadata->taxonomy = $taxonomy;
449	}
450
451	/**
452	* Filters the post metadata sent to the BeyondWords API.
453	*
454	* Scheduled for removal in plugin version 5.0.0.
455	*
456	* @since 3.3.0
457	*
458	* @deprecated 4.3.0 Set the 'metadata' key in beyondwords_content_params instead.
459	*
460	* @param object $metadata Post metadata. Defaults to the taxonomies and terms assigned to the post.
461	* @param int $postId Post ID.
462	*/
463	$metadata = apply_filters('beyondwords_post_metadata', $metadata, $postId);
464
465	return $metadata;
466	}
467
468	/**
469	* Get all taxonomies, and their selected terms, for a post.
470	*
471	* Returns an associative array of taxonomy names and terms.
472	*
473	* For example:
474	*
475	* array(
476	* "categories" => array("Category 1"),
477	* "post_tag" => array("Tag 1", "Tag 2", "Tag 3"),
478	* )
479	*
480	* @since 3.3.0
481	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
482	*
483	* @param int $postId Post ID.
484	*
485	* @return array
486	*/
487	public static function getAllTaxonomiesAndTerms($postId)
488	{
489	$postType = get_post_type($postId);
490
491	$postTypeTaxonomies = get_object_taxonomies($postType);
492
493	$taxonomies = new \stdClass();
494
495	foreach ($postTypeTaxonomies as $postTypeTaxonomy) {
496	$terms = get_the_terms($postId, $postTypeTaxonomy);
497
498	if (! empty($terms) && ! is_wp_error($terms)) {
499	$taxonomies->{(string)$postTypeTaxonomy} = wp_list_pluck($terms, 'name');
500	}
501	}
502
503	return $taxonomies;
504	}
505
506	/**
507	* Get author name for a post.
508	*
509	* @since 3.10.4
510	*
511	* @param int $postId Post ID.
512	*
513	* @return string
514	*/
515	public static function getAuthorName($postId)
516	{
517	$authorId = get_post_field('post_author', $postId);
518
519	return get_the_author_meta('display_name', $authorId);
520	}
521
522	/**
523	* Add data-beyondwords-marker attribute to the root elements in a HTML
524	* string (typically the rendered HTML of a single block).
525	*
526	* Checks to see whether we can use WP_HTML_Tag_Processor, or whether we
527	* fall back to using DOMDocument to add the marker.
528	*
529	* @since 4.2.2
530	*
531	* @param string $html HTML.
532	* @param string $marker Marker UUID.
533	*
534	* @return string HTML.
535	*/
536	public static function addMarkerAttribute($html, $marker)
537	{
538	if (! $marker) {
539	return $html;
540	}
541
542	// Prefer WP_HTML_Tag_Processor, introduced in WordPress 6.2
543	if (class_exists('WP_HTML_Tag_Processor')) {
544	return PostContentUtils::addMarkerAttributeWithHTMLTagProcessor($html, $marker);
545	} else {
546	return PostContentUtils::addMarkerAttributeWithDOMDocument($html, $marker);
547	}
548	}
549
550	/**
551	* Add data-beyondwords-marker attribute to the root elements in a HTML
552	* string using WP_HTML_Tag_Processor.
553	*
554	* @since 4.0.0
555	* @since 4.2.2 Moved from src/Component/Post/BlockAttributes/BlockAttributes.php
556	* to src/Component/Post/PostContentUtils.php
557	* @since 4.7.0 Prevent empty data-beyondwords-marker attributes.
558	*
559	* @param string $html HTML.
560	* @param string $marker Marker UUID.
561	*
562	* @return string HTML.
563	*/
564	public static function addMarkerAttributeWithHTMLTagProcessor($html, $marker)
565	{
566	if (! $marker) {
567	return $html;
568	}
569
570	// https://github.com/WordPress/gutenberg/pull/42485
571	$tags = new \WP_HTML_Tag_Processor($html);
572
573	if ($tags->next_tag()) {
574	$tags->set_attribute('data-beyondwords-marker', $marker);
575	}
576
577	return strval($tags);
578	}
579
580	/**
581	* Add data-beyondwords-marker attribute to the root elements in a HTML
582	* string using DOMDocument.
583	*
584	* This is a fallback, since WP_HTML_Tag_Processor was only shipped with
585	* WordPress 6.2 on 19 April 2023.
586	*
587	* https://make.wordpress.org/core/2022/10/13/whats-new-in-gutenberg-14-3-12-october/
588	*
589	* Note: It is not ideal to do all the $bodyElement/$fullHtml processing
590	* in this method, but without it DOMDocument does not work as expected if
591	* there is more than 1 root element. The approach here has been taken from
592	* some historic Gutenberg code before they implemented WP_HTML_Tag_Processor:
593	*
594	* https://github.com/WordPress/gutenberg/blob/6671cef1179412a2bbd4969cbbc82705c7f69bac/lib/block-supports/index.php
595	*
596	* @since 4.0.0
597	* @since 4.2.2 Moved from src/Component/Post/BlockAttributes/BlockAttributes.php
598	* to src/Component/Post/PostContentUtils.php
599	* @since 4.7.0 Prevent empty data-beyondwords-marker attributes.
600	*
601	* @param string $html HTML.
602	* @param string $marker Marker UUID.
603	*
604	* @return string HTML.
605	*/
606	public static function addMarkerAttributeWithDOMDocument($html, $marker)
607	{
608	if (! $marker) {
609	return $html;
610	}
611
612	$dom = new \DOMDocument('1.0', 'utf-8');
613
614	$wrappedHtml =
615	'<html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></head><body>'
616	. $html
617	. '</body></html>';
618
619	$success = $dom->loadHTML($wrappedHtml, LIBXML_HTML_NODEFDTD \| LIBXML_COMPACT);
620
621	if (! $success) {
622	return $html;
623	}
624
625	// Structure is like `<html><head/><body/></html>`, so body is the `lastChild` of our document.
626	$bodyElement = $dom->documentElement->lastChild;
627
628	$xpath = new \DOMXPath($dom);
629	$blockRoot = $xpath->query('./*', $bodyElement)[0];
630
631	if (empty($blockRoot)) {
632	return $html;
633	}
634
635	$blockRoot->setAttribute('data-beyondwords-marker', $marker);
636
637	// Avoid using `$dom->saveHtml( $node )` because the node results may not produce consistent
638	// whitespace. Saving the root HTML `$dom->saveHtml()` prevents this behavior.
639	$fullHtml = $dom->saveHtml();
640
641	// Find the <body> open/close tags. The open tag needs to be adjusted so we get inside the tag
642	// and not the tag itself.
643	$start = strpos($fullHtml, '<body>', 0) + strlen('<body>');
644	$end = strpos($fullHtml, '</body>', $start);
645
646	return trim(substr($fullHtml, $start, $end - $start));
647	}
648	}
649