speechkit / src / Component / Post / PostContentUtils.php

PostContentUtils.php in BeyondWords – AI audio for publishers 4.5.1, at src/Component/Post/PostContentUtils.php

608 lines 19.8 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 <[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 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	// Trim to remove trailing newlines – common for WordPress content
100	$content = trim($content);
101
102	/**
103	* Filters the content body we send for audio processing.
104	*
105	* Scheduled for removal in plugin version 5.0.0.
106	*
107	* @since 4.0.0
108	*
109	* @deprecated 4.3.0 Set the 'body' key in beyondwords_content_params instead.
110	*
111	* @param string $content The post content.
112	* @param int $postId The post ID.
113	*/
114	$content = apply_filters('beyondwords_content', $content, $post->ID);
115
116	return $content;
117	}
118
119	/**
120	* Get the summary for the audio content, ready to be sent to the BeyondWords API.
121	*
122	* @param int\|WP_Post $post The WordPress post ID, or post object.
123	*
124	* @since 4.0.0
125	*
126	* @return string The summary.
127	*/
128	public static function getSummary($post)
129	{
130	$post = get_post($post);
131
132	if (!($post instanceof \WP_Post)) {
133	throw new \Exception('Post Not Found');
134	}
135
136	$summary = null;
137
138	// Optionally send the excerpt to the REST API, if the plugin setting has been checked
139	$prependExcerpt = get_option('beyondwords_prepend_excerpt');
140
141	if ($prependExcerpt && has_excerpt($post)) {
142	// Escape characters
143	$summary = htmlentities($post->post_excerpt, ENT_QUOTES \| ENT_XHTML);
144	// Apply WordPress filters
145	$summary = apply_filters('get_the_excerpt', $summary);
146	// Convert line breaks into paragraphs
147	$summary = trim(wpautop($summary));
148	}
149
150	return $summary;
151	}
152
153	/**
154	* Get the segments for the audio content, ready to be sent to the BeyondWords API.
155	*
156	* THIS METHOD IS CURRENTLY NOT IN USE. Segments cannot currently include HTML
157	* formatting tags such as <strong> and <em> so we do not pass segments, we pass
158	* a HTML string as the body param instead.
159	*
160	* @param int\|WP_Post $post The WordPress post ID, or post object.
161	*
162	* @since 4.0.0
163	*
164	* @return array\|null The segments.
165	*/
166	public static function getSegments($post)
167	{
168	if (! has_blocks($post)) {
169	return null;
170	}
171
172	$titleSegment = (object) [
173	'section' => 'title',
174	'text' => get_the_title($post),
175	];
176
177	$summarySegment = (object) [
178	'section' => 'summary',
179	'text' => PostContentUtils::getSummary($post),
180	];
181
182	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
183
184	$bodySegments = array_map(function ($block) {
185	$marker = null;
186
187	if (isset($block['attrs']) && isset($block['attrs']['beyondwordsMarker'])) {
188	$marker = $block['attrs']['beyondwordsMarker'];
189	}
190
191	return (object) [
192	'section' => 'body',
193	'marker' => $marker,
194	'text' => trim(render_block($block)),
195	];
196	}, $blocks);
197
198	// Merge title, summary and body segments
199	$segments = array_values(array_merge([$titleSegment], [$summarySegment], $bodySegments));
200
201	// Remove any segments with empty text
202	$segments = array_values(array_filter($segments, function ($segment) {
203	return (! empty($segment->text));
204	}));
205
206	return $segments;
207	}
208
209	/**
210	* Get the post content without blocks which have been filtered.
211	*
212	* We have added buttons into the Gutenberg editor to optionally exclude selected
213	* blocks from the source text for audio.
214	*
215	* This method filters all blocks, removing any which have been excluded.
216	*
217	* @param int\|WP_Post $post The WordPress post ID, or post object.
218	*
219	* @since 3.8.0
220	* @since 4.0.0 Replace for loop with array_reduce
221	*
222	* @return string The post body without excluded blocks.
223	*/
224	public static function getContentWithoutExcludedBlocks($post)
225	{
226	if (! has_blocks($post)) {
227	return trim($post->post_content);
228	}
229
230	$blocks = parse_blocks($post->post_content);
231	$output = '';
232
233	$blocks = PostContentUtils::getAudioEnabledBlocks($post);
234
235	foreach ($blocks as $block) {
236	$marker = $block['attrs']['beyondwordsMarker'] ?? '';
237
238	$output .= PostContentUtils::addMarkerAttribute(
239	render_block($block),
240	$marker
241	);
242	}
243
244	return $output;
245	}
246
247	/**
248	* Get audio-enabled blocks.
249	*
250	* @param int\|WP_Post $post The WordPress post ID, or post object.
251	*
252	* @since 4.0.0
253	*
254	* @return array The blocks.
255	*/
256	public static function getAudioEnabledBlocks($post)
257	{
258	$post = get_post($post);
259
260	if (! ($post instanceof \WP_Post)) {
261	return [];
262	}
263
264	if (! has_blocks($post)) {
265	return [];
266	}
267
268	$allBlocks = parse_blocks($post->post_content);
269
270	$blocks = array_filter($allBlocks, function ($block) {
271	$enabled = true;
272
273	if (is_array($block['attrs']) && isset($block['attrs']['beyondwordsAudio'])) {
274	$enabled = (bool) $block['attrs']['beyondwordsAudio'];
275	}
276
277	return $enabled;
278	});
279
280	/**
281	* Filters the audio-enabled blocks for a post.
282	*
283	* Scheduled for removal in plugin version 5.0.0.
284	*
285	* @since 4.0.0
286	*
287	* @deprecated 4.3.0 Replace with {@link https://docs.beyondwords.io/docs-and-guides/content/filter-content}.
288	*
289	* @param array $blocks The audio-enabled post blocks.
290	* @param array $allBlocks All post blocks including those with audio disabled.
291	* @param int $postId The post ID.
292	*/
293	$blocks = apply_filters('beyondwords_post_audio_enabled_blocks', $blocks, $allBlocks, $post->ID);
294
295	return $blocks;
296	}
297
298	/**
299	* Get the body param we pass to the API.
300	*
301	* @since 3.0.0 Introduced as getBodyJson.
302	* @since 3.3.0 Added metadata to aid custom playlist generation.
303	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils.
304	* @since 3.10.4 Rename `published_at` API param to `publish_date`.
305	* @since 4.0.0 Use new API params.
306	* @since 4.0.3 Ensure `image_url` is always a string.
307	* @since 4.3.0 Rename from getBodyJson to getContentParams.
308	*
309	* @static
310	* @param int $postId WordPress Post ID.
311	*
312	* @return Response
313	**/
314	public static function getContentParams($postId)
315	{
316	$body = [
317	'type' => 'auto_segment',
318	'title' => get_the_title($postId),
319	'summary' => PostContentUtils::getSummary($postId),
320	'body' => PostContentUtils::getBody($postId),
321	'source_url' => get_the_permalink($postId),
322	'source_id' => strval($postId),
323	'author' => PostContentUtils::getAuthorName($postId),
324	'image_url' => strval(wp_get_original_image_url(get_post_thumbnail_id($postId))),
325	'metadata' => PostContentUtils::getMetadata($postId),
326	'published' => true,
327	'publish_date' => get_post_time(PostContentUtils::DATE_FORMAT, true, $postId),
328	];
329
330	$status = get_post_status($postId);
331
332	/*
333	* If the post status is "pending" then we send { published: false } to
334	* the BeyondWords API, to prevent the generated audio from being
335	* published in playlists.
336	*
337	* We also omit { publish_date } because get_post_time() returns `false`
338	* for posts which are "Pending Review".
339	*/
340	if ($status === 'pending') {
341	$body['published'] = false;
342	unset($body['publish_date']);
343	}
344
345	$bodyVoiceId = intval(get_post_meta($postId, 'beyondwords_body_voice_id', true));
346
347	if ($bodyVoiceId > 0) {
348	$body['body_voice_id'] = $bodyVoiceId;
349	}
350
351	$titleVoiceId = intval(get_post_meta($postId, 'beyondwords_title_voice_id', true));
352
353	if ($titleVoiceId > 0) {
354	$body['title_voice_id'] = $titleVoiceId;
355	}
356
357	$summaryVoiceId = intval(get_post_meta($postId, 'beyondwords_summary_voice_id', true));
358
359	if ($summaryVoiceId > 0) {
360	$body['summary_voice_id'] = $summaryVoiceId;
361	}
362
363	/**
364	* Filters the params we send to the BeyondWords API 'content' endpoint.
365	*
366	* Scheduled for removal in plugin version 5.0.0.
367	*
368	* @since 4.0.0
369	*
370	* @deprecated 4.3.0 Replaced with beyondwords_content_params.
371	*
372	* @param array $body The params we send to the BeyondWords API.
373	* @param array $postId WordPress post ID.
374	*/
375	$body = apply_filters('beyondwords_body_params', $body, $postId);
376
377	/**
378	* Filters the params we send to the BeyondWords API 'content' endpoint.
379	*
380	* @since 4.0.0 Introduced as beyondwords_body_params
381	* @since 4.3.0 Renamed from beyondwords_body_params to beyondwords_content_params
382	*
383	* @param array $body The params we send to the BeyondWords API.
384	* @param array $postId WordPress post ID.
385	*/
386	$body = apply_filters('beyondwords_content_params', $body, $postId);
387
388	return wp_json_encode($body);
389	}
390
391	/**
392	* Get the post metadata to send with BeyondWords API requests.
393	*
394	* The metadata key is defined by the BeyondWords API as "A custom object
395	* for storing meta information".
396	*
397	* The metadata values are used to create filters for playlists in the
398	* BeyondWords dashboard.
399	*
400	* We currently only include taxonomies by default, and the output of this
401	* method can be filtered using the `beyondwords_post_metadata` filter.
402	*
403	* @since 3.3.0
404	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
405	*
406	* @param int $postId Post ID.
407	*
408	* @return array
409	*/
410	public static function getMetadata($postId)
411	{
412	$metadata = new \stdClass();
413
414	$taxonomy = PostContentUtils::getAllTaxonomiesAndTerms($postId);
415
416	if (count((array)$taxonomy)) {
417	$metadata->taxonomy = $taxonomy;
418	}
419
420	/**
421	* Filters the post metadata sent to the BeyondWords API.
422	*
423	* Scheduled for removal in plugin version 5.0.0.
424	*
425	* @since 3.3.0
426	*
427	* @deprecated 4.3.0 Set the 'metadata' key in beyondwords_content_params instead.
428	*
429	* @param object $metadata Post metadata. Defaults to the taxonomies and terms assigned to the post.
430	* @param int $postId Post ID.
431	*/
432	$metadata = apply_filters('beyondwords_post_metadata', $metadata, $postId);
433
434	return $metadata;
435	}
436
437	/**
438	* Get all taxonomies, and their selected terms, for a post.
439	*
440	* Returns an associative array of taxonomy names and terms.
441	*
442	* For example:
443	*
444	* array(
445	* "categories" => array("Category 1"),
446	* "post_tag" => array("Tag 1", "Tag 2", "Tag 3"),
447	* )
448	*
449	* @since 3.3.0
450	* @since 3.5.0 Moved from Core\Utils to Component\Post\PostUtils
451	*
452	* @param int $postId Post ID.
453	*
454	* @return array
455	*/
456	public static function getAllTaxonomiesAndTerms($postId)
457	{
458	$postType = get_post_type($postId);
459
460	$postTypeTaxonomies = get_object_taxonomies($postType);
461
462	$taxonomies = new \stdClass();
463
464	foreach ($postTypeTaxonomies as $postTypeTaxonomy) {
465	$terms = get_the_terms($postId, $postTypeTaxonomy);
466
467	if (! empty($terms) && ! is_wp_error($terms)) {
468	$taxonomies->{(string)$postTypeTaxonomy} = wp_list_pluck($terms, 'name');
469	}
470	}
471
472	return $taxonomies;
473	}
474
475	/**
476	* Get author name for a post.
477	*
478	* @since 3.10.4
479	*
480	* @param int $postId Post ID.
481	*
482	* @return string
483	*/
484	public static function getAuthorName($postId)
485	{
486	$authorId = get_post_field('post_author', $postId);
487
488	return get_the_author_meta('display_name', $authorId);
489	}
490
491	/**
492	* Add data-beyondwords-marker attribute to the root elements in a HTML
493	* string (typically the rendered HTML of a single block).
494	*
495	* Checks to see whether we can use WP_HTML_Tag_Processor, or whether we
496	* fall back to using DOMDocument to add the marker.
497	*
498	* @since 4.2.2
499	*
500	* @param string $html HTML.
501	* @param string $marker Marker UUID.
502	*
503	* @return string HTML.
504	*/
505	public static function addMarkerAttribute($html, $marker)
506	{
507	if (! $marker) {
508	return $html;
509	}
510
511	// Prefer WP_HTML_Tag_Processor, introduced in WordPress 6.2
512	if (class_exists('WP_HTML_Tag_Processor')) {
513	return PostContentUtils::addMarkerAttributeWithHTMLTagProcessor($html, $marker);
514	} else {
515	return PostContentUtils::addMarkerAttributeWithDOMDocument($html, $marker);
516	}
517	}
518
519	/**
520	* Add data-beyondwords-marker attribute to the root elements in a HTML
521	* string using WP_HTML_Tag_Processor.
522	*
523	* @since 4.0.0
524	* @since 4.2.2 Moved from src/Component/Post/BlockAttributes/BlockAttributes.php
525	* to src/Component/Post/PostContentUtils.php
526	*
527	* @param string $html HTML.
528	* @param string $marker Marker UUID.
529	*
530	* @return string HTML.
531	*/
532	public static function addMarkerAttributeWithHTMLTagProcessor($html, $marker)
533	{
534	// https://github.com/WordPress/gutenberg/pull/42485
535	$tags = new \WP_HTML_Tag_Processor($html);
536
537	if ($tags->next_tag()) {
538	$tags->set_attribute('data-beyondwords-marker', $marker);
539	}
540
541	return strval($tags);
542	}
543
544	/**
545	* Add data-beyondwords-marker attribute to the root elements in a HTML
546	* string using DOMDocument.
547	*
548	* This is a fallback, since WP_HTML_Tag_Processor was only shipped with
549	* WordPress 6.2 on 19 April 2023.
550	*
551	* https://make.wordpress.org/core/2022/10/13/whats-new-in-gutenberg-14-3-12-october/
552	*
553	* Note: It is not ideal to do all the $bodyElement/$fullHtml processing
554	* in this method, but without it DOMDocument does not work as expected if
555	* there is more than 1 root element. The approach here has been taken from
556	* some historic Gutenberg code before they implemented WP_HTML_Tag_Processor:
557	*
558	* https://github.com/WordPress/gutenberg/blob/6671cef1179412a2bbd4969cbbc82705c7f69bac/lib/block-supports/index.php
559	*
560	* @since 4.0.0
561	* @since 4.2.2 Moved from src/Component/Post/BlockAttributes/BlockAttributes.php
562	* to src/Component/Post/PostContentUtils.php
563	*
564	* @param string $html HTML.
565	* @param string $marker Marker UUID.
566	*
567	* @return string HTML.
568	*/
569	public static function addMarkerAttributeWithDOMDocument($html, $marker)
570	{
571	$dom = new \DOMDocument('1.0', 'utf-8');
572
573	$wrappedHtml =
574	'<html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></head><body>'
575	. $html
576	. '</body></html>';
577
578	$success = $dom->loadHTML($wrappedHtml, LIBXML_HTML_NODEFDTD \| LIBXML_COMPACT);
579
580	if (! $success) {
581	return $html;
582	}
583
584	// Structure is like `<html><head/><body/></html>`, so body is the `lastChild` of our document.
585	$bodyElement = $dom->documentElement->lastChild;
586
587	$xpath = new \DOMXPath($dom);
588	$blockRoot = $xpath->query('./*', $bodyElement)[0];
589
590	if (empty($blockRoot)) {
591	return $html;
592	}
593
594	$blockRoot->setAttribute('data-beyondwords-marker', $marker);
595
596	// Avoid using `$dom->saveHtml( $node )` because the node results may not produce consistent
597	// whitespace. Saving the root HTML `$dom->saveHtml()` prevents this behavior.
598	$fullHtml = $dom->saveHtml();
599
600	// Find the <body> open/close tags. The open tag needs to be adjusted so we get inside the tag
601	// and not the tag itself.
602	$start = strpos($fullHtml, '<body>', 0) + strlen('<body>');
603	$end = strpos($fullHtml, '</body>', $start);
604
605	return trim(substr($fullHtml, $start, $end - $start));
606	}
607	}
608