activitypub / includes / class-hashtag.php

class-hashtag.php in ActivityPub trunk, at includes/class-hashtag.php

159 lines 4.8 KB

Raw Download Zip

1	<?php
2	/**
3	* Hashtag Class.
4	*
5	* @package Activitypub
6	*/
7
8	namespace Activitypub;
9
10	/**
11	* ActivityPub Hashtag Class.
12	*
13	* @author Matthias Pfefferle
14	*/
15	class Hashtag {
16	/**
17	* Initialize the class, registering WordPress hooks.
18	*/
19	public static function init() {
20	if ( '1' === \get_option( 'activitypub_use_hashtags', '0' ) ) {
21	\add_action( 'wp_insert_post', array( self::class, 'insert_post' ), 10, 2 );
22	\add_filter( 'the_content', array( self::class, 'the_content' ) );
23	\add_filter( 'activitypub_activity_object_array', array( self::class, 'filter_activity_object' ), 99 );
24	}
25	}
26
27	/**
28	* Filter only the activity object and replace summery it with URLs.
29	*
30	* @param array $activity The activity object array.
31	*
32	* @return array The filtered activity object array.
33	*/
34	public static function filter_activity_object( $activity ) {
35	if ( ! empty( $activity['summary'] ) && is_actor( $activity ) ) {
36	$activity['summary'] = self::the_content( $activity['summary'] );
37	}
38
39	if ( ! empty( $activity['content'] ) ) {
40	$activity['content'] = self::the_content( $activity['content'] );
41	}
42
43	return $activity;
44	}
45
46	/**
47	* Filter to save #tags as real WordPress tags.
48	*
49	* @param int $post_id Post ID.
50	* @param \WP_Post $post Post object.
51	*/
52	public static function insert_post( $post_id, $post ) {
53	// Check if the post supports ActivityPub.
54	if ( ! \post_type_supports( \get_post_type( $post ), 'activitypub' ) ) {
55	return;
56	}
57
58	// Check if the (custom) post supports tags.
59	$taxonomies = \get_object_taxonomies( $post );
60	if ( ! \in_array( 'post_tag', $taxonomies, true ) ) {
61	return;
62	}
63
64	$content = $post->post_content . "\n" . $post->post_excerpt;
65	$content = self::extract_text_outside_protected_tags( $content );
66
67	$tags = array();
68	if ( \preg_match_all( '/' . ACTIVITYPUB_HASHTAGS_REGEXP . '/i', $content, $match ) ) {
69	$tags = \array_unique( $match[1] );
70	}
71
72	\wp_add_post_tags( $post->ID, \implode( ', ', $tags ) );
73	}
74
75	/**
76	* Extract text content from outside protected HTML elements.
77	*
78	* Uses WP_HTML_Tag_Processor to properly parse HTML and skip content inside
79	* protected tags, matching the behavior of enrich_content_data().
80	*
81	* @param string $content The HTML content to process.
82	*
83	* @return string Text content from non-protected areas only.
84	*/
85	private static function extract_text_outside_protected_tags( $content ) {
86	$processor = new \WP_HTML_Tag_Processor( $content );
87
88	/*
89	* Do not process content inside protected tags.
90	*
91	* Note: SCRIPT, STYLE, and TEXTAREA are "atomic" elements in
92	* WP_HTML_Tag_Processor, meaning their content is bundled with the tag
93	* token and won't appear as separate #text nodes. Because of this they
94	* do not need to be listed in $protected_tags: their inner text is
95	* never surfaced as #text tokens for us to process.
96	* See https://github.com/WordPress/wordpress-develop/blob/0fb3bb29596918864d808d156268a2df63c83620/src/wp-includes/html-api/class-wp-html-tag-processor.php#L276
97	*/
98	$protected_tags = array( 'PRE', 'CODE', 'A' );
99	$tag_stack = array();
100	$filtered_content = '';
101
102	while ( $processor->next_token() ) {
103	$token_type = $processor->get_token_type();
104
105	if ( '#tag' === $token_type ) {
106	$tag_name = $processor->get_tag();
107
108	if ( $processor->is_tag_closer() ) {
109	// Closing tag: remove from stack.
110	$i = \array_search( $tag_name, $tag_stack, true );
111	if ( false !== $i ) {
112	$tag_stack = \array_slice( $tag_stack, 0, $i );
113	}
114	} elseif ( \in_array( $tag_name, $protected_tags, true ) ) {
115	// Opening tag: add to stack.
116	$tag_stack[] = $tag_name;
117	}
118	} elseif ( '#text' === $token_type && empty( $tag_stack ) ) {
119	// Only include text chunks that are outside protected tags.
120	$filtered_content .= $processor->get_modifiable_text();
121	}
122	}
123
124	return $filtered_content;
125	}
126
127	/**
128	* Filter to replace the #tags in the content with links.
129	*
130	* @param string $the_content The post content.
131	*
132	* @return string The filtered post content.
133	*/
134	public static function the_content( $the_content ) {
135	return enrich_content_data( $the_content, '/' . ACTIVITYPUB_HASHTAGS_REGEXP . '/i', array( self::class, 'replace_with_links' ) );
136	}
137
138	/**
139	* A callback for preg_replace to build the term links.
140	*
141	* @param array $result The preg_match results.
142	* @return string the final string
143	*/
144	public static function replace_with_links( $result ) {
145	$tag = $result[1];
146	$tag_object = \get_term_by( 'name', $tag, 'post_tag' );
147	if ( ! $tag_object ) {
148	$tag_object = \get_term_by( 'name', $tag, 'category' );
149	}
150
151	if ( $tag_object ) {
152	$link = \get_term_link( $tag_object, 'post_tag' );
153	return \sprintf( '<a rel="tag" class="hashtag u-tag u-category" href="%s">#%s</a>', \esc_url( $link ), $tag );
154	}
155
156	return '#' . $tag;
157	}
158	}
159