activitypub / includes / class-blocklist-subscriptions.php

class-blocklist-subscriptions.php in ActivityPub trunk, at includes/class-blocklist-subscriptions.php

223 lines 5.3 KB

Raw Download Zip

1	<?php
2	/**
3	* Blocklist Subscriptions class file.
4	*
5	* @package Activitypub
6	*/
7
8	namespace Activitypub;
9
10	/**
11	* Blocklist Subscriptions class.
12	*
13	* Manages subscriptions to remote blocklists for automatic updates.
14	* Owns all remote blocklist logic: fetching, parsing, and importing.
15	*/
16	class Blocklist_Subscriptions {
17
18	/**
19	* Option key for storing subscriptions.
20	*/
21	const OPTION_KEY = 'activitypub_blocklist_subscriptions';
22
23	/**
24	* IFTAS DNI list URL.
25	*/
26	const IFTAS_DNI_URL = 'https://about.iftas.org/wp-content/uploads/2025/10/iftas-dni-latest.csv';
27
28	/**
29	* Get all subscriptions.
30	*
31	* @return array Array of URL => timestamp pairs.
32	*/
33	public static function get_all() {
34	return \get_option( self::OPTION_KEY, array() );
35	}
36
37	/**
38	* Add a subscription.
39	*
40	* Only adds the URL to the subscription list. Does not sync.
41	* Call sync() separately to fetch and import domains.
42	*
43	* @param string $url The blocklist URL to subscribe to.
44	* @return bool True on success, false on failure.
45	*/
46	public static function add( $url ) {
47	$url = \sanitize_url( $url );
48
49	if ( empty( $url ) \|\| ! \filter_var( $url, FILTER_VALIDATE_URL ) ) {
50	return false;
51	}
52
53	$subscriptions = self::get_all();
54
55	// Not already subscribed.
56	if ( ! isset( $subscriptions[ $url ] ) ) {
57	// Add subscription with timestamp 0 (never synced).
58	$subscriptions[ $url ] = 0;
59	\update_option( self::OPTION_KEY, $subscriptions );
60	}
61
62	return true;
63	}
64
65	/**
66	* Remove a subscription.
67	*
68	* @param string $url The blocklist URL to unsubscribe from.
69	* @return bool True on success, false if not found.
70	*/
71	public static function remove( $url ) {
72	$subscriptions = self::get_all();
73
74	if ( ! isset( $subscriptions[ $url ] ) ) {
75	return false;
76	}
77
78	unset( $subscriptions[ $url ] );
79	\update_option( self::OPTION_KEY, $subscriptions );
80
81	return true;
82	}
83
84	/**
85	* Sync a single subscription.
86	*
87	* Fetches the blocklist URL, parses domains, and adds new ones to the blocklist.
88	* Updates the subscription timestamp on success.
89	*
90	* @param string $url The blocklist URL to sync.
91	* @return int\|false Number of domains added, or false on failure.
92	*/
93	public static function sync( $url ) {
94	$response = \wp_safe_remote_get(
95	$url,
96	array(
97	'timeout' => 30,
98	'redirection' => 5,
99	)
100	);
101
102	if ( \is_wp_error( $response ) ) {
103	return false;
104	}
105
106	$response_code = \wp_remote_retrieve_response_code( $response );
107	if ( 200 !== $response_code ) {
108	return false;
109	}
110
111	$body = \wp_remote_retrieve_body( $response );
112	if ( empty( $body ) ) {
113	return false;
114	}
115
116	$domains = self::parse_csv_string( $body );
117
118	if ( empty( $domains ) ) {
119	return false;
120	}
121
122	// Get existing blocks and find new ones.
123	$existing = Moderation::get_site_blocks()[ Moderation::TYPE_DOMAIN ] ?? array();
124	$new_domains = \array_diff( $domains, $existing );
125
126	if ( ! empty( $new_domains ) ) {
127	Moderation::add_site_blocks( Moderation::TYPE_DOMAIN, $new_domains );
128	}
129
130	// Update timestamp if this is a subscription.
131	$subscriptions = self::get_all();
132	if ( isset( $subscriptions[ $url ] ) ) {
133	$subscriptions[ $url ] = \time();
134	\update_option( self::OPTION_KEY, $subscriptions );
135	}
136
137	return \count( $new_domains );
138	}
139
140	/**
141	* Sync all subscriptions.
142	*
143	* Called by cron job.
144	*/
145	public static function sync_all() {
146	\array_map( array( __CLASS__, 'sync' ), \array_keys( self::get_all() ) );
147	}
148
149	/**
150	* Parse CSV content from a string and extract domain names.
151	*
152	* Supports Mastodon CSV format (with #domain header) and simple
153	* one-domain-per-line format.
154	*
155	* @param string $content CSV content as a string.
156	* @return array Array of unique, valid domain names.
157	*/
158	public static function parse_csv_string( $content ) {
159	$domains = array();
160
161	if ( empty( $content ) ) {
162	return $domains;
163	}
164
165	// Split into lines.
166	$lines = \preg_split( '/\r\n\|\r\|\n/', $content );
167	if ( empty( $lines ) ) {
168	return $domains;
169	}
170
171	// Parse first line to detect format.
172	$first_line = \str_getcsv( $lines[0], ',', '"', '\\' );
173	$first_cell = \trim( $first_line[0] ?? '' );
174	$has_header = \str_starts_with( $first_cell, '#' ) \|\| 'domain' === \strtolower( $first_cell );
175
176	// Find domain column index.
177	$domain_index = 0;
178	if ( $has_header ) {
179	foreach ( $first_line as $i => $col ) {
180	$col = \ltrim( \strtolower( \trim( $col ) ), '#' );
181	if ( 'domain' === $col ) {
182	$domain_index = $i;
183	break;
184	}
185	}
186	// Remove header from lines.
187	\array_shift( $lines );
188	}
189
190	// Process each line.
191	foreach ( $lines as $line ) {
192	$row = \str_getcsv( $line, ',', '"', '\\' );
193	$domain = \trim( $row[ $domain_index ] ?? '' );
194
195	// Skip empty lines and comments.
196	if ( empty( $domain ) \|\| \str_starts_with( $domain, '#' ) ) {
197	continue;
198	}
199
200	if ( self::is_valid_domain( $domain ) ) {
201	$domains[] = \strtolower( $domain );
202	}
203	}
204
205	return \array_unique( $domains );
206	}
207
208	/**
209	* Validate a domain name.
210	*
211	* @param string $domain The domain to validate.
212	* @return bool True if valid, false otherwise.
213	*/
214	public static function is_valid_domain( $domain ) {
215	// Must contain at least one dot (filter_var would accept "localhost").
216	if ( ! \str_contains( $domain, '.' ) ) {
217	return false;
218	}
219
220	return (bool) \filter_var( $domain, FILTER_VALIDATE_DOMAIN, FILTER_FLAG_HOSTNAME );
221	}
222	}
223