wp-content-filter / includes / class-filter-engine.php

class-filter-engine.php in Content Censor trunk, at includes/class-filter-engine.php

220 lines 6.0 KB

Raw Download Zip

1	<?php
2	/**
3	* Deterministic content filtering engine.
4	*
5	* @package WPGO_Content_Censor
6	*/
7
8	namespace WPGO\Content_Censor;
9
10	if ( ! defined( 'ABSPATH' ) ) {
11	exit;
12	}
13
14	/**
15	* Filters scalar strings and string arrays without depending on global state.
16	*/
17	final class Filter_Engine {
18
19	/** Maximum number of configured keywords evaluated in one pass. */
20	const MAX_KEYWORDS = 500;
21
22	/** Maximum UTF-8 length of one keyword. */
23	const MAX_KEYWORD_LENGTH = 200;
24
25	/**
26	* Filter a WordPress value.
27	*
28	* @param mixed $value String or array of strings.
29	* @param array<string, mixed> $options Plugin options.
30	* @param int $post_id Current post ID, when available.
31	* @return mixed
32	*/
33	public static function filter( $value, array $options, $post_id = 0 ) {
34	if ( self::is_excluded( (int) $post_id, $options ) ) {
35	return $value;
36	}
37
38	if ( is_array( $value ) ) {
39	$filtered = array();
40	foreach ( $value as $key => $item ) {
41	$filtered[ $key ] = self::filter( $item, $options, $post_id );
42	}
43	return $filtered;
44	}
45
46	if ( ! is_string( $value ) \|\| '' === $value ) {
47	return $value;
48	}
49
50	$keywords = self::keywords( $options );
51	if ( array() === $keywords ) {
52	return $value;
53	}
54
55	if ( '1' === (string) ( $options['chk_ignore_html_attr'] ?? '0' ) && function_exists( 'wp_html_split' ) ) {
56	$parts = wp_html_split( $value );
57	foreach ( $parts as $index => $part ) {
58	if ( '' !== $part && '<' !== $part[0] ) {
59	$parts[ $index ] = self::filter_text( $part, $keywords, $options );
60	}
61	}
62	return implode( '', $parts );
63	}
64
65	return self::filter_text( $value, $keywords, $options );
66	}
67
68	/**
69	* Parse and normalize configured keywords.
70	*
71	* @param array<string, mixed> $options Plugin options.
72	* @return array<int, string>
73	*/
74	public static function keywords( array $options ) {
75	$delimiter_map = array(
76	'pipe' => '\|',
77	'comma' => ',',
78	'pound' => '#',
79	);
80	$delimiter = $delimiter_map[ $options['drp_keyword_sep'] ?? 'pipe' ] ?? '\|';
81	$raw = (string) ( $options['txtar_keywords'] ?? '' );
82	$keywords = array();
83
84	foreach ( explode( $delimiter, $raw ) as $candidate ) {
85	$candidate = trim( $candidate );
86	$length = self::length( $candidate );
87	if ( $length < 3 \|\| $length > self::MAX_KEYWORD_LENGTH ) {
88	continue;
89	}
90	$keywords[ $candidate ] = $candidate;
91	if ( count( $keywords ) >= self::MAX_KEYWORDS ) {
92	break;
93	}
94	}
95
96	$keywords = array_values( $keywords );
97	usort(
98	$keywords,
99	static function ( $left, $right ) {
100	return Filter_Engine::length( $right ) <=> Filter_Engine::length( $left );
101	}
102	);
103	return $keywords;
104	}
105
106	/**
107	* Render one matched word using the selected mask.
108	*
109	* @param string $word Matched word.
110	* @param string $mode all, first, or firstlast.
111	* @param string $wildcard Replacement character.
112	* @return string
113	*/
114	public static function censor_word( $word, $mode, $wildcard ) {
115	$characters = preg_split( '//u', $word, -1, PREG_SPLIT_NO_EMPTY );
116	if ( false === $characters ) {
117	$characters = str_split( $word );
118	}
119	$count = count( $characters );
120	if ( 0 === $count ) {
121	return $word;
122	}
123
124	if ( 'first' === $mode ) {
125	return $characters[0] . str_repeat( $wildcard, max( 0, $count - 1 ) );
126	}
127	if ( 'firstlast' === $mode && $count > 1 ) {
128	return $characters[0] . str_repeat( $wildcard, max( 0, $count - 2 ) ) . $characters[ $count - 1 ];
129	}
130	return str_repeat( $wildcard, $count );
131	}
132
133	/**
134	* Filter one plain-text fragment.
135	*
136	* @param string $text Text fragment.
137	* @param array<int, string> $keywords Normalized keywords.
138	* @param array<string, mixed> $options Plugin options.
139	* @return string
140	*/
141	private static function filter_text( $text, array $keywords, array $options ) {
142	$alternatives = array_map(
143	static function ( $keyword ) {
144	return preg_quote( $keyword, '/' );
145	},
146	$keywords
147	);
148	$body = '(?:' . implode( '\|', $alternatives ) . ')';
149	$strict = 'strict_off' !== (string) ( $options['rdo_strict_filtering'] ?? 'strict_on' );
150	if ( $strict ) {
151	$body = '(?<![\p{L}\p{N}_])' . $body . '(?![\p{L}\p{N}_])';
152	}
153
154	$case_insensitive = 'sen' !== (string) ( $options['rdo_case'] ?? 'insen' );
155	$utf8 = 1 === preg_match( '//u', $text );
156	$modifiers = ( $case_insensitive ? 'i' : '' ) . ( $utf8 ? 'u' : '' );
157	$pattern = '/' . $body . '/' . $modifiers;
158	$mode = (string) ( $options['rdo_word'] ?? 'all' );
159	$wildcard = self::wildcard( (string) ( $options['drp_filter_char'] ?? 'star' ) );
160
161	$result = preg_replace_callback(
162	$pattern,
163	static function ( array $matched ) use ( $mode, $wildcard ) {
164	return Filter_Engine::censor_word( $matched[0], $mode, $wildcard );
165	},
166	$text
167	);
168
169	return null === $result ? $text : $result;
170	}
171
172	/**
173	* Check the configured exclusion list.
174	*
175	* @param int $post_id Current post ID.
176	* @param array<string, mixed> $options Plugin options.
177	* @return bool
178	*/
179	private static function is_excluded( $post_id, array $options ) {
180	if ( $post_id <= 0 ) {
181	return false;
182	}
183	$ids = preg_split( '/\s,\s/', (string) ( $options['txt_exclude'] ?? '' ), -1, PREG_SPLIT_NO_EMPTY );
184	if ( false === $ids ) {
185	return false;
186	}
187	return in_array( $post_id, array_map( 'intval', $ids ), true );
188	}
189
190	/**
191	* Convert the setting key into a replacement character.
192	*
193	* @param string $name Setting key.
194	* @return string
195	*/
196	private static function wildcard( $name ) {
197	$wildcards = array(
198	'star' => '*',
199	'dollar' => '$',
200	'question' => '?',
201	'exclamation' => '!',
202	'hyphen' => '-',
203	'hash' => '#',
204	'tilde' => '~',
205	'blank' => '',
206	);
207	return $wildcards[ $name ] ?? '*';
208	}
209
210	/**
211	* Return the Unicode-aware string length when available.
212	*
213	* @param string $value String value.
214	* @return int
215	*/
216	private static function length( $value ) {
217	return function_exists( 'mb_strlen' ) ? mb_strlen( $value, 'UTF-8' ) : strlen( $value );
218	}
219	}
220