easyfonts / src / Frontend / HtmlProcessor.php

HtmlProcessor.php in EasyFonts – Host Google Fonts Locally, Fast & Auto-Optimize, GDPR Compliant 2.0.0, at src/Frontend/HtmlProcessor.php

230 lines 5.9 KB

Raw Download Zip

1	<?php
2	/**
3	* HTML manipulation backed by WordPress's native HTML API.
4	*
5	* Link discovery/rewriting goes through WP_HTML_Tag_Processor (a real
6	* tokenizer — no catastrophic backtracking). Operations the tokenizer can't do
7	* in core (reading <style> text, deleting nodes) use tightly-scoped helpers.
8	*
9	* @package EasyFonts
10	*/
11
12	namespace EasyFonts\Frontend;
13
14	use WP_HTML_Tag_Processor;
15
16	defined( 'ABSPATH' ) \|\| exit;
17
18	/**
19	* Reliable HTML read/rewrite helpers.
20	*/
21	class HtmlProcessor {
22
23	/**
24	* Iterate every <link rel="stylesheet">, letting a callback rewrite the href.
25	*
26	* @param string $html HTML.
27	* @param callable $callback fn(string $href, array $attrs): ?string — return new href or null to skip.
28	* @return array{html:string,changed:bool}
29	*/
30	public function rewrite_stylesheet_links( string $html, callable $callback ): array {
31	$processor = new WP_HTML_Tag_Processor( $html );
32	$changed = false;
33
34	while ( $processor->next_tag( array( 'tag_name' => 'LINK' ) ) ) {
35	$rel = (string) $processor->get_attribute( 'rel' );
36
37	if ( false === stripos( $rel, 'stylesheet' ) ) {
38	continue;
39	}
40
41	$href = $processor->get_attribute( 'href' );
42
43	if ( ! is_string( $href ) \|\| '' === $href ) {
44	continue;
45	}
46
47	$attrs = array(
48	'id' => $processor->get_attribute( 'id' ),
49	'media' => $processor->get_attribute( 'media' ),
50	);
51
52	$new = $callback( $href, $attrs );
53
54	if ( is_string( $new ) && $new !== $href ) {
55	$processor->set_attribute( 'href', $new );
56	$changed = true;
57	}
58	}
59
60	return array(
61	'html' => $processor->get_updated_html(),
62	'changed' => $changed,
63	);
64	}
65
66	/**
67	* Collect all <link rel="stylesheet"> hrefs (read-only).
68	*
69	* @param string $html HTML.
70	* @return string[]
71	*/
72	public function stylesheet_hrefs( string $html ): array {
73	$processor = new WP_HTML_Tag_Processor( $html );
74	$hrefs = array();
75
76	while ( $processor->next_tag( array( 'tag_name' => 'LINK' ) ) ) {
77	$rel = (string) $processor->get_attribute( 'rel' );
78
79	if ( false === stripos( $rel, 'stylesheet' ) ) {
80	continue;
81	}
82
83	$href = $processor->get_attribute( 'href' );
84
85	if ( is_string( $href ) && '' !== $href ) {
86	$hrefs[] = $href;
87	}
88	}
89
90	return $hrefs;
91	}
92
93	/**
94	* Physically remove specific <link> stylesheet tags by exact href.
95	*
96	* The tokenizer can't delete nodes, so we mark them inert with the
97	* tokenizer (reliable matching) and then strip the marked tags by regex.
98	*
99	* @param string $html HTML.
100	* @param string[] $hrefs Exact href values to remove.
101	* @return string
102	*/
103	public function remove_stylesheet_links( string $html, array $hrefs ): string {
104	if ( empty( $hrefs ) ) {
105	return $html;
106	}
107
108	$targets = array_fill_keys( $hrefs, true );
109	$processor = new WP_HTML_Tag_Processor( $html );
110	$hit = false;
111
112	while ( $processor->next_tag( array( 'tag_name' => 'LINK' ) ) ) {
113	$rel = (string) $processor->get_attribute( 'rel' );
114
115	if ( false === stripos( $rel, 'stylesheet' ) ) {
116	continue;
117	}
118
119	$href = $processor->get_attribute( 'href' );
120
121	if ( is_string( $href ) && isset( $targets[ $href ] ) ) {
122	$processor->set_attribute( 'data-easyfonts-remove', '1' );
123	$hit = true;
124	}
125	}
126
127	if ( ! $hit ) {
128	return $html;
129	}
130
131	$html = $processor->get_updated_html();
132
133	// Drop any <link …> carrying our marker.
134	$html = preg_replace( '#<link\b[^>]\bdata-easyfonts-remove=([\'"])1\1[^>]/?>\s*#i', '', $html );
135
136	return null === $html ? '' : $html;
137	}
138
139	/**
140	* Process the CSS contents of each <style> block.
141	*
142	* @param string $html HTML.
143	* @param callable $callback fn(string $css): ?string — return new CSS or null to leave unchanged.
144	* @return array{html:string,changed:bool}
145	*/
146	public function rewrite_style_blocks( string $html, callable $callback ): array {
147	$changed = false;
148
149	$out = preg_replace_callback(
150	'/(<style\b[^>]>)(.?)(<\/style>)/is',
151	static function ( $m ) use ( $callback, &$changed ) {
152	$new = $callback( $m[2] );
153
154	if ( is_string( $new ) && $new !== $m[2] ) {
155	$changed = true;
156	return $m[1] . $new . $m[3];
157	}
158
159	return $m[0];
160	},
161	$html
162	);
163
164	return array(
165	'html' => null === $out ? $html : $out,
166	'changed' => $changed,
167	);
168	}
169
170	/**
171	* Remove Google-font resource hints (preconnect / dns-prefetch / preload).
172	*
173	* @param string $html HTML.
174	* @param string[] $domains Domains to strip hints for.
175	* @return array{html:string,changed:bool}
176	*/
177	public function strip_font_hints( string $html, array $domains ): array {
178	$processor = new WP_HTML_Tag_Processor( $html );
179	$remove = array();
180
181	// Identify offending hint tags by their href (read pass).
182	while ( $processor->next_tag( array( 'tag_name' => 'LINK' ) ) ) {
183	$rel = strtolower( (string) $processor->get_attribute( 'rel' ) );
184
185	if ( ! in_array( $rel, array( 'preconnect', 'dns-prefetch', 'preload' ), true ) ) {
186	continue;
187	}
188
189	$href = (string) $processor->get_attribute( 'href' );
190
191	foreach ( $domains as $domain ) {
192	if ( false !== strpos( $href, $domain ) ) {
193	// Neutralise: blank the href + rel so the hint is inert.
194	$processor->set_attribute( 'href', '' );
195	$processor->set_attribute( 'rel', 'easyfonts-removed' );
196	$remove[] = true;
197	break;
198	}
199	}
200	}
201
202	$html = $processor->get_updated_html();
203
204	// Physically drop the neutralised tags.
205	$html = preg_replace( '/<link\b[^>]\brel=([\'"])easyfonts-removed\1[^>]>\s*/i', '', $html );
206
207	return array(
208	'html' => null === $html ? '' : $html,
209	'changed' => ! empty( $remove ),
210	);
211	}
212
213	/**
214	* Insert a snippet immediately before </head>.
215	*
216	* @param string $html HTML.
217	* @param string $snippet Snippet.
218	* @return string
219	*/
220	public function before_head_close( string $html, string $snippet ): string {
221	$pos = stripos( $html, '</head>' );
222
223	if ( false === $pos ) {
224	return $html;
225	}
226
227	return substr_replace( $html, $snippet . "\n</head>", $pos, 7 );
228	}
229	}
230