speechkit / src / Component / Post / PostContentUtils.php

PostContentUtils.php in BeyondWords – AI audio for publishers 4.6.0, at src/Component/Post/PostContentUtils.php

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