speechkit / src / Component / Post / PostContentUtils.php

PostContentUtils.php in BeyondWords – AI audio for publishers 4.4.0, at src/Component/Post/PostContentUtils.php

612 lines 20.1 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 source text for the audio, ready to be sent to the BeyondWords API.
21	*
22	* @deprecated 4.0.0 Renamed to PostContentUtils::getBody()
23	*
24	* @param int\|WP_Post $post The WordPress post ID, or post object.
25	*
26	* @since 3.0.0
27	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
28	* @since 3.8.0 Exclude Gutenberg blocks with attribute { beyondwordsAudio: false }
29	* @since 4.0.0 Renamed from PostContentUtils::getSourceTextForAudio() to PostContentUtils::getBody()
30	*
31	* @return string The body (the processed $post->post_content).
32	*/
33	public static function getSourceTextForAudio($post)
34	{
35	_doing_it_wrong(
36	'PostContentUtils::getBody',
37	'BeyondWords PostContentUtils::getSourceTextForAudio() has been renamed to PostContentUtils::getBody()',
38	'4.0.0'
39	);
40
41	return PostContentUtils::getBody($post);
42	}
43
44	/**
45	* Get the body for the audio, ready to be sent to the BeyondWords API.
46	*
47	* The following rules are applied:
48	*
49	* Main body content entered in WordPress
50	* + Optionally filtered using [SpeechKit-Start]/[SpeechKit-Stop] "shortcodes"
51	* + With registered content filters FROM OTHER PLUGINS applied
52	* + Optionally prepended with the Post excerpt
53	* + Optionally filtered using the beyondwords_content filter
54	*
55	* @SuppressWarnings(PHPMD.LongVariable)
56	*
57	* @param int\|WP_Post $post The WordPress post ID, or post object.
58	*
59	* @since 3.0.0
60	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
61	* @since 3.8.0 Exclude Gutenberg blocks with attribute { beyondwordsAudio: false }
62	* @since 4.0.0 Renamed from PostContentUtils::getSourceTextForAudio() to PostContentUtils::getBody()
63	*
64	* @return string The body (the processed $post->post_content).
65	*/
66	public static function getBody($post)
67	{
68	global $beyondwords_wordpress_plugin;
69
70	$post = get_post($post);
71
72	if (!($post instanceof \WP_Post)) {
73	throw new \Exception('Post Not Found');
74	}
75
76	$content = PostContentUtils::getContentWithoutExcludedBlocks($post);
77
78	// If SpeechKit-Start/Stop tags are present then use the content within them
79	// @deprecated v3.0.0: publishers should use the beyondwords_content filter instead.
80	$regex = '/\[SpeechKit-Start\](.*?)\[SpeechKit-Stop\]/s';
81
82	if (preg_match_all($regex, $content, $match, PREG_PATTERN_ORDER) > 0) {
83	$content = implode(' ', $match[1]);
84	}
85
86	// Temporarily remove our Player filter, to exclude the player <div>
87	if ($beyondwords_wordpress_plugin && isset($beyondwords_wordpress_plugin->player)) {
88	remove_filter('the_content', array($beyondwords_wordpress_plugin->player, 'autoPrependPlayer'));
89	}
90
91	// Apply other standard WordPress filters to handle shortcodes etc
92	$content = apply_filters('the_content', $content);
93
94	// Add our Player filter back in again
95	if ($beyondwords_wordpress_plugin && isset($beyondwords_wordpress_plugin->player)) {
96	add_filter('the_content', array($beyondwords_wordpress_plugin->player, 'autoPrependPlayer'));
97	}
98
99	// TODO maybe we need to do something like this to deal with Shortcodes a little better?
100	// $content = wp_strip_all_tags(apply_filters('the_content', $content));
101
102	// Trim to remove trailing newlines – common for WordPress content
103	$content = trim($content);
104
105	/**
106	* Filters the content body we send for audio processing.
107	*
108	* Scheduled for removal in plugin version 5.0.0.
109	*
110	* @since 4.0.0
111	*
112	* @deprecated 4.3.0 Set the 'body' key in beyondwords_content_params instead.
113	*
114	* @param string $content The post content.
115	* @param int $postId The post ID.
116	*/
117	$content = apply_filters('beyondwords_content', $content, $post->ID);
118
119	return $content;
120	}
121
122	/**
123	* Get the summary for the audio content, ready to be sent to the BeyondWords API.
124	*
125	* @param int\|WP_Post $post The WordPress post ID, or post object.
126	*
127	* @since 4.0.0
128	*
129	* @return string The summary.
130	*/
131	public static function getSummary($post)
132	{
133	$post = get_post($post);
134
135	if (!($post instanceof \WP_Post)) {
136	throw new \Exception('Post Not Found');
137	}
138
139	$summary = null;
140
141	// Optionally send the excerpt to the REST API, if the plugin setting has been checked
142	$prependExcerpt = get_option('beyondwords_prepend_excerpt');
143
144	if ($prependExcerpt && has_excerpt($post)) {
145	// Escape characters
146	$summary = htmlentities($post->post_excerpt, ENT_QUOTES \| ENT_XHTML);
147	// Apply WordPress filters
148	$summary = apply_filters('get_the_excerpt', $summary);
149	// Convert line breaks into paragraphs
150	$summary = trim(wpautop($summary));
151	}
152
153	return $summary;
154	}
155
156	/**
157	* Get the segments for the audio content, ready to be sent to the BeyondWords API.
158	*
159	* THIS METHOD IS CURRENTLY NOT IN USE. Segments cannot currently include HTML
160	* formatting tags such as <strong> and <em> so we do not pass segments, we pass
161	* a HTML string as the body param instead.
162	*
163	* @param int\|WP_Post $post The WordPress post ID, or post object.
164	*
165	* @since 4.0.0
166	*
167	* @return array\|null The segments.
168	*/
169	public static function getSegments($post)
170	{
171	if (! has_blocks($post)) {
172	return null;
173	}
174
175	$titleSegment = (object) [
176	'section' => 'title',
177	'text' => get_the_title($post),
178	];
179
180	$summarySegment = (object) [
181	'section' => 'summary',
182	'text' => PostContentUtils::getSummary($post),
183	];
184
185	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
186
187	$bodySegments = array_map(function ($block) {
188	$marker = null;
189
190	if (isset($block['attrs']) && isset($block['attrs']['beyondwordsMarker'])) {
191	$marker = $block['attrs']['beyondwordsMarker'];
192	}
193
194	return (object) [
195	'section' => 'body',
196	'marker' => $marker,
197	'text' => trim(render_block($block)),
198	];
199	}, $blocks);
200
201	// Merge title, summary and body segments
202	$segments = array_values(array_merge([$titleSegment], [$summarySegment], $bodySegments));
203
204	// TODO Consider removing this when API can handle it
205	// Remove any segments with empty text
206	$segments = array_values(array_filter($segments, function ($segment) {
207	return (! empty($segment->text));
208	}));
209
210	return $segments;
211	}
212
213	/**
214	* Get the post content without blocks which have been filtered.
215	*
216	* We have added buttons into the Gutenberg editor to optionally exclude selected
217	* blocks from the source text for audio.
218	*
219	* This method filters all blocks, removing any which have been excluded.
220	*
221	* @param int\|WP_Post $post The WordPress post ID, or post object.
222	*
223	* @since 3.8.0
224	* @since 4.0.0 Replace for loop with array_reduce
225	*
226	* @return string The post body without excluded blocks.
227	*/
228	public static function getContentWithoutExcludedBlocks($post)
229	{
230	if (! has_blocks($post)) {
231	return trim($post->post_content);
232	}
233
234	$blocks = parse_blocks($post->post_content);
235	$output = '';
236
237	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
238
239	foreach ($blocks as $block) {
240	$marker = $block['attrs']['beyondwordsMarker'] ?? '';
241
242	$output .= PostContentUtils::addMarkerAttribute(
243	render_block($block),
244	$marker
245	);
246	}
247
248	return $output;
249	}
250
251	/**
252	* Get audio-enabled blocks.
253	*
254	* @param int\|WP_Post $post The WordPress post ID, or post object.
255	*
256	* @since 4.0.0
257	*
258	* @return array The blocks.
259	*/
260	public static function getAudioEnabledBlocks($post)
261	{
262	$post = get_post($post);
263
264	if (! ($post instanceof \WP_Post)) {
265	return [];
266	}
267
268	if (! has_blocks($post)) {
269	return [];
270	}
271
272	$allBlocks = parse_blocks($post->post_content);
273
274	$blocks = array_filter($allBlocks, function ($block) {
275	$enabled = true;
276
277	if (is_array($block['attrs']) && isset($block['attrs']['beyondwordsAudio'])) {
278	$enabled = (bool) $block['attrs']['beyondwordsAudio'];
279	}
280
281	return $enabled;
282	});
283
284	/**
285	* Filters the audio-enabled blocks for a post.
286	*
287	* Scheduled for removal in plugin version 5.0.0.
288	*
289	* @since 4.0.0
290	*
291	* @deprecated 4.3.0 Replace with {@link https://docs.beyondwords.io/docs-and-guides/content/filter-content}.
292	*
293	* @param array $blocks The audio-enabled post blocks.
294	* @param array $allBlocks All post blocks including those with audio disabled.
295	* @param int $postId The post ID.
296	*/
297	$blocks = apply_filters('beyondwords_post_audio_enabled_blocks', $blocks, $allBlocks, $post->ID);
298
299	return $blocks;
300	}
301
302	/**
303	* Get the body param we pass to the API.
304	*
305	* @since 3.0.0 Introduced as getBodyJson.
306	* @since 3.3.0 Added metadata to aid custom playlist generation.
307	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils.
308	* @since 3.10.4 Rename `published_at` API param to `publish_date`.
309	* @since 4.0.0 Use new API params.
310	* @since 4.0.3 Ensure `image_url` is always a string.
311	* @since 4.3.0 Rename from getBodyJson to getContentParams.
312	*
313	* @static
314	* @param int $postId WordPress Post ID.
315	*
316	* @return Response
317	**/
318	public static function getContentParams($postId)
319	{
320	$body = [
321	'type' => 'auto_segment',
322	'title' => get_the_title($postId),
323	'summary' => PostContentUtils::getSummary($postId),
324	'body' => PostContentUtils::getBody($postId),
325	'source_url' => get_the_permalink($postId),
326	'source_id' => strval($postId),
327	'author' => PostContentUtils::getAuthorName($postId),
328	'image_url' => strval(wp_get_original_image_url(get_post_thumbnail_id($postId))),
329	'metadata' => PostContentUtils::getMetadata($postId),
330	'published' => true,
331	'publish_date' => get_post_time(PostContentUtils::DATE_FORMAT, true, $postId),
332	];
333
334	$status = get_post_status($postId);
335
336	/*
337	* If the post status is "pending" then we send { published: false } to
338	* the BeyondWords API, to prevent the generated audio from being
339	* published in playlists.
340	*
341	* We also omit { publish_date } because get_post_time() returns `false`
342	* for posts which are "Pending Review".
343	*/
344	if ($status === 'pending') {
345	$body['published'] = false;
346	unset($body['publish_date']);
347	}
348
349	$bodyVoiceId = intval(get_post_meta($postId, 'beyondwords_body_voice_id', true));
350
351	if ($bodyVoiceId > 0) {
352	$body['body_voice_id'] = $bodyVoiceId;
353	}
354
355	$titleVoiceId = intval(get_post_meta($postId, 'beyondwords_title_voice_id', true));
356
357	if ($titleVoiceId > 0) {
358	$body['title_voice_id'] = $titleVoiceId;
359	}
360
361	$summaryVoiceId = intval(get_post_meta($postId, 'beyondwords_summary_voice_id', true));
362
363	if ($summaryVoiceId > 0) {
364	$body['summary_voice_id'] = $summaryVoiceId;
365	}
366
367	/**
368	* Filters the params we send to the BeyondWords API 'content' endpoint.
369	*
370	* Scheduled for removal in plugin version 5.0.0.
371	*
372	* @since 4.0.0
373	*
374	* @deprecated 4.3.0 Replaced with beyondwords_content_params.
375	*
376	* @param array $body The params we send to the BeyondWords API.
377	* @param array $postId WordPress post ID.
378	*/
379	$body = apply_filters('beyondwords_body_params', $body, $postId);
380
381	/**
382	* Filters the params we send to the BeyondWords API 'content' endpoint.
383	*
384	* @since 4.0.0 Introduced as beyondwords_body_params
385	* @since 4.3.0 Renamed from beyondwords_body_params to beyondwords_content_params
386	*
387	* @param array $body The params we send to the BeyondWords API.
388	* @param array $postId WordPress post ID.
389	*/
390	$body = apply_filters('beyondwords_content_params', $body, $postId);
391
392	return wp_json_encode($body);
393	}
394
395	/**
396	* Get the post metadata to send with BeyondWords API requests.
397	*
398	* The metadata key is defined by the BeyondWords API as "A custom object
399	* for storing meta information".
400	*
401	* The metadata values are used to create filters for playlists in the
402	* BeyondWords dashboard.
403	*
404	* We currently only include taxonomies by default, and the output of this
405	* method can be filtered using the `beyondwords_post_metadata` filter.
406	*
407	* @since 3.3.0
408	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
409	*
410	* @param int $postId Post ID.
411	*
412	* @return array
413	*/
414	public static function getMetadata($postId)
415	{
416	$metadata = new \stdClass();
417
418	$taxonomy = PostContentUtils::getAllTaxonomiesAndTerms($postId);
419
420	if (count((array)$taxonomy)) {
421	$metadata->taxonomy = $taxonomy;
422	}
423
424	/**
425	* Filters the post metadata sent to the BeyondWords API.
426	*
427	* Scheduled for removal in plugin version 5.0.0.
428	*
429	* @since 3.3.0
430	*
431	* @deprecated 4.3.0 Set the 'metadata' key in beyondwords_content_params instead.
432	*
433	* @param object $metadata Post metadata. Defaults to the taxonomies and terms assigned to the post.
434	* @param int $postId Post ID.
435	*/
436	$metadata = apply_filters('beyondwords_post_metadata', $metadata, $postId);
437
438	return $metadata;
439	}
440
441	/**
442	* Get all taxonomies, and their selected terms, for a post.
443	*
444	* Returns an associative array of taxonomy names and terms.
445	*
446	* For example:
447	*
448	* array(
449	* "categories" => array("Category 1"),
450	* "post_tag" => array("Tag 1", "Tag 2", "Tag 3"),
451	* )
452	*
453	* @since 3.3.0
454	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
455	*
456	* @param int $postId Post ID.
457	*
458	* @return array
459	*/
460	public static function getAllTaxonomiesAndTerms($postId)
461	{
462	$postType = get_post_type($postId);
463
464	$postTypeTaxonomies = get_object_taxonomies($postType);
465
466	$taxonomies = new \stdClass();
467
468	foreach ($postTypeTaxonomies as $postTypeTaxonomy) {
469	$terms = get_the_terms($postId, $postTypeTaxonomy);
470
471	if (! empty($terms) && ! is_wp_error($terms)) {
472	$taxonomies->{(string)$postTypeTaxonomy} = wp_list_pluck($terms, 'name');
473	}
474	}
475
476	return $taxonomies;
477	}
478
479	/**
480	* Get author name for a post.
481	*
482	* @since 3.10.4
483	*
484	* @param int $postId Post ID.
485	*
486	* @return string
487	*/
488	public static function getAuthorName($postId)
489	{
490	$authorId = get_post_field('post_author', $postId);
491
492	return get_the_author_meta('display_name', $authorId);
493	}
494
495	/**
496	* Add data-beyondwords-marker attribute to the root elements in a HTML
497	* string (typically the rendered HTML of a single block).
498	*
499	* Checks to see whether we can use WP_HTML_Tag_Processor, or whether we
500	* fall back to using DOMDocument to add the marker.
501	*
502	* @since 4.2.2
503	*
504	* @param string $html HTML.
505	* @param string $marker Marker UUID.
506	*
507	* @return string HTML.
508	*/
509	public static function addMarkerAttribute($html, $marker)
510	{
511	if (! $marker) {
512	return $html;
513	}
514
515	// Prefer WP_HTML_Tag_Processor, introduced in WordPress 6.2
516	if (class_exists('WP_HTML_Tag_Processor')) {
517	return PostContentUtils::addMarkerAttributeWithHTMLTagProcessor($html, $marker);
518	} else {
519	return PostContentUtils::addMarkerAttributeWithDOMDocument($html, $marker);
520	}
521	}
522
523	/**
524	* Add data-beyondwords-marker attribute to the root elements in a HTML
525	* string using WP_HTML_Tag_Processor.
526	*
527	* @since 4.0.0
528	* @since 4.2.2 Moved from src/Component/Post/BlockAttributes/BlockAttributes.php
529	* to src/Component/Post/PostContentUtils.php
530	*
531	* @param string $html HTML.
532	* @param string $marker Marker UUID.
533	*
534	* @return string HTML.
535	*/
536	public static function addMarkerAttributeWithHTMLTagProcessor($html, $marker)
537	{
538	// https://github.com/WordPress/gutenberg/pull/42485
539	$tags = new \WP_HTML_Tag_Processor($html);
540
541	if ($tags->next_tag()) {
542	$tags->set_attribute('data-beyondwords-marker', $marker);
543	}
544
545	return strval($tags);
546	}
547
548	/**
549	* Add data-beyondwords-marker attribute to the root elements in a HTML
550	* string using DOMDocument.
551	*
552	* This is a fallback, since WP_HTML_Tag_Processor was only shipped with
553	* WordPress 6.2 on 19 April 2023.
554	*
555	* https://make.wordpress.org/core/2022/10/13/whats-new-in-gutenberg-14-3-12-october/
556	*
557	* Note: It is not ideal to do all the $bodyElement/$fullHtml processing
558	* in this method, but without it DOMDocument does not work as expected if
559	* there is more than 1 root element. The approach here has been taken from
560	* some historic Gutenberg code before they implemented WP_HTML_Tag_Processor:
561	*
562	* https://github.com/WordPress/gutenberg/blob/6671cef1179412a2bbd4969cbbc82705c7f69bac/lib/block-supports/index.php
563	*
564	* @since 4.0.0
565	* @since 4.2.2 Moved from src/Component/Post/BlockAttributes/BlockAttributes.php
566	* to src/Component/Post/PostContentUtils.php
567	*
568	* @param string $html HTML.
569	* @param string $marker Marker UUID.
570	*
571	* @return string HTML.
572	*/
573	public static function addMarkerAttributeWithDOMDocument($html, $marker)
574	{
575	$dom = new \DOMDocument('1.0', 'utf-8');
576
577	$wrappedHtml =
578	'<html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></head><body>'
579	. $html
580	. '</body></html>';
581
582	$success = $dom->loadHTML($wrappedHtml, LIBXML_HTML_NODEFDTD \| LIBXML_COMPACT);
583
584	if (! $success) {
585	return $html;
586	}
587
588	// Structure is like `<html><head/><body/></html>`, so body is the `lastChild` of our document.
589	$bodyElement = $dom->documentElement->lastChild;
590
591	$xpath = new \DOMXPath($dom);
592	$blockRoot = $xpath->query('./*', $bodyElement)[0];
593
594	if (empty($blockRoot)) {
595	return $html;
596	}
597
598	$blockRoot->setAttribute('data-beyondwords-marker', $marker);
599
600	// Avoid using `$dom->saveHtml( $node )` because the node results may not produce consistent
601	// whitespace. Saving the root HTML `$dom->saveHtml()` prevents this behavior.
602	$fullHtml = $dom->saveHtml();
603
604	// Find the <body> open/close tags. The open tag needs to be adjusted so we get inside the tag
605	// and not the tag itself.
606	$start = strpos($fullHtml, '<body>', 0) + strlen('<body>');
607	$end = strpos($fullHtml, '</body>', $start);
608
609	return trim(substr($fullHtml, $start, $end - $start));
610	}
611	}
612