autoptimize / classes / autoptimizeAttributeParser.php

autoptimizeAttributeParser.php in Autoptimize 3.1.16, at classes/autoptimizeAttributeParser.php

259 lines 8.4 KB

Raw Download Zip

1	<?php
2
3	if ( ! defined( 'ABSPATH' ) ) {
4	exit;
5	}
6
7	class autoptimizeAttributeParser
8	{
9	/**
10	* Protocols allowed during attribute parsing.
11	* Extends WP's default list with protocols that legitimately appear in HTML
12	* attributes but would otherwise be stripped by wp_kses_hair.
13	*/
14	private static function allowed_protocols(): array {
15	return array_merge( wp_allowed_protocols(), [ 'data', 'blob' ] );
16	}
17
18	/**
19	* Entry point for modifying HTML attributes.
20	*/
21	public static function modify( $tag, $target, $swaps = [], $replacements = []): string
22	{
23	$pieces = wp_html_split( $tag );
24
25	foreach ( $pieces as &$piece ) {
26	// Skip empty strings, text nodes, and comments.
27	if ( $piece === '' \|\| $piece[0] !== '<' \|\| $piece[1] === '!' ) {
28	continue;
29	}
30
31	// Skip closing tags.
32	if ( $piece[1] === '/' ) {
33	continue;
34	}
35
36	// Extract tag name; skip if not parseable.
37	if ( ! preg_match( '/^<([a-z0-9]+)/i', $piece, $tag_name_match ) ) {
38	continue;
39	}
40
41	// Skip non-matching tags.
42	if ( strcasecmp( $tag_name_match[1], $target ) !== 0 ) {
43	continue;
44	}
45
46	// This piece matches $target — apply modifications.
47	$attr_string = preg_replace( '/^<[a-z0-9]+\s*\|(\s?\/?>)$/i', '', $piece );
48	$attrs = wp_kses_hair( $attr_string, self::allowed_protocols() );
49
50	// 1. Handle Deletions first to clear the deck
51	foreach ( $replacements as $attr_name => $new_value ) {
52	if ( is_null( $new_value ) ) {
53	unset( $attrs[$attr_name] );
54	}
55	}
56
57	// 2. Handle Swaps
58	foreach ( $swaps as $old => $new ) {
59	if ( isset( $attrs[$old] ) ) {
60	$attrs[$new] = $attrs[$old];
61	$attrs[$new]['name'] = $new;
62	unset( $attrs[$old] );
63	}
64	}
65
66	// 3. Handle Replacements & Additions
67	foreach ( $replacements as $attr_name => $new_value ) {
68	if ( ! is_null( $new_value ) ) {
69	$attrs[$attr_name] = [
70	'name' => $attr_name,
71	'value' => $new_value,
72	];
73	}
74	}
75
76	// Rebuild this tag.
77	$rebuilt = '<' . $tag_name_match[1];
78	foreach ( $attrs as $name => $data ) {
79	$value = esc_attr( $data['value'] );
80	$rebuilt .= " {$name}=\"{$value}\"";
81	}
82	$rebuilt .= ( strpos( $piece, '/>' ) !== false ) ? ' />' : '>';
83
84	$piece = $rebuilt;
85	}
86	unset( $piece );
87
88	return implode( '', $pieces );
89	}
90
91	/**
92	* Entry point for retrieving attributes.
93	* Returns a single value (string\|null) if a string name is provided,
94	* or an associative array if a regex is provided.
95	*/
96	public static function get( $tag, $target, $query ) // : string\|array\|null TODO: uncomment when PHP7.4 is no longer supported
97	{
98	$pieces = wp_html_split( $tag );
99
100	foreach ( $pieces as $piece ) {
101	// Skip empty strings, text nodes, and comments.
102	if ( $piece === '' \|\| $piece[0] !== '<' \|\| $piece[1] === '!' ) {
103	continue;
104	}
105
106	// Skip closing tags.
107	if ( $piece[1] === '/' ) {
108	continue;
109	}
110
111	// Extract tag name.
112	if ( ! preg_match( '/^<([a-z0-9]+)/i', $piece, $tag_name_match ) ) {
113	continue;
114	}
115
116	// Skip non-matching tags.
117	if ( strcasecmp( $tag_name_match[1], $target ) !== 0 ) {
118	continue;
119	}
120
121	// Found a matching tag — parse its attributes.
122	$attr_string = preg_replace( '/^<[a-z0-9]+\s*\|(\s?\/?>)$/i', '', $piece );
123	$attrs = wp_kses_hair( $attr_string, self::allowed_protocols() );
124
125	// Single attribute request
126	if ( is_string( $query ) && strpos( $query, '/' ) !== 0 ) {
127	return $attrs[$query]['value'] ?? null;
128	}
129
130	// Regex request
131	$results = [];
132	foreach ( $attrs as $name => $data ) {
133	if ( preg_match( $query, $name ) ) {
134	$results[$name] = $data['value'];
135	}
136	}
137	return $results;
138	}
139
140	// No matching tag found.
141	return is_string( $query ) && strpos( $query, '/' ) !== 0 ? null : [];
142	}
143
144	/**
145	* Helper to rename attributes.
146	* Supports:
147	* - rename( $tag, 'src', 'data-src' )
148	* - rename( $tag, ['src' => 'data-src', 'srcset' => 'data-srcset'] )
149	*/
150	public static function rename( $tag, $target, $old_name, $new_name = null ): string {
151	if ( is_array( $old_name ) ) {
152	$swaps = $old_name;
153	} else {
154	$swaps = [ $old_name => $new_name ];
155	}
156
157	return self::modify( $tag, $target, $swaps );
158	}
159
160	/**
161	* Helper to remove one or more attributes.
162	* Supports:
163	* - remove( $tag, 'style' )
164	* - remove( $tag, ['style', 'onclick', 'onerror'] )
165	*/
166	public static function remove( $tag, $target, $attributes ): string {
167	// array_fill_keys expects an array; (array) cast handles single strings safely.
168	$replacements = array_fill_keys( (array) $attributes, null );
169
170	return self::modify( $tag, $target, [], $replacements );
171	}
172
173	/**
174	* Helper to replace or add attribute values.
175	* Supports:
176	* - replace( $tag, 'class', 'lazyload' )
177	* - replace( $tag, ['class' => 'lazyload', 'src' => 'placeholder.png'] )
178	*/
179	public static function replace( $tag, $target, $attribute, $value = null ): string {
180	if ( is_array( $attribute ) ) {
181	$replacements = $attribute;
182	} else {
183	$replacements = [ $attribute => $value ];
184	}
185
186	return self::modify( $tag, $target, [], $replacements );
187	}
188
189	/**
190	* Helper to rebuild a HTML tag from a string
191	* Can be used to fix minified HTML that is missing quotes around attributes
192	* Supports:
193	* - replace( $tag )
194	*/
195	public static function rebuild( $tag, $target ): string
196	{
197	return self::modify( $tag, $target, [], [], true );
198	}
199
200	/**
201	* Retrieves a single attribute value.
202	* @param string $tag The HTML tag string.
203	* @param string $name The attribute name to find.
204	* @return string\|null The attribute value or null if not found.
205	*/
206	public static function get_attribute( string $tag, string $target, string $name ): ?string
207	{
208	$result = self::get( $tag, $target, $name );
209	return is_string( $result ) ? $result : null;
210	}
211
212	/**
213	* Retrieves a list of attributes matching a regex or prefix.
214	* @param string $tag The HTML tag string.
215	* @param string $regex The regex pattern to match keys against (e.g., '/^data-/').
216	* @return array Associative array of [name => value].
217	*/
218	public static function get_attributes( string $tag, string $target, string $regex ): array
219	{
220	$result = self::get( $tag, $target, $regex );
221	return is_array( $result ) ? $result : [];
222	}
223
224	/**
225	* Modifies a single specific attribute via callback.
226	* If callback returns null, the attribute is removed.
227	*/
228	public static function modify_attribute( string $tag, string $target, string $name, callable $callback ): string
229	{
230	$value = self::get_attribute( $tag, $target, $name );
231	$new_value = $callback( $name, $value );
232
233	if ( $new_value === $value ) {
234	return $tag;
235	}
236
237	return self::modify( $tag, $target, [], [ $name => $new_value ] );
238	}
239
240	/**
241	* Modifies multiple attributes matching a regex via callback.
242	*/
243	public static function modify_attributes( string $tag, string $target, string $regex, callable $callback ): string
244	{
245	$attrs = self::get_attributes( $tag, $target, $regex );
246	$replacements = [];
247
248	foreach ( $attrs as $name => $value ) {
249	$new_value = $callback( $name, $value );
250
251	if ( $new_value !== $value ) {
252	$replacements[ $name ] = $new_value;
253	}
254	}
255
256	return empty( $replacements ) ? $tag : self::modify( $tag, $target, [], $replacements );
257	}
258	}
259