| 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 |
* Is WordPress's HTML API available? (WP 6.2+.) When it isn't, every method |
| 25 |
* here fails open — returning the HTML unchanged — so a missing core class can |
| 26 |
* never fatal inside the output buffer and white-screen the site. |
| 27 |
* |
| 28 |
* @return bool |
| 29 |
*/ |
| 30 |
private function api_ready(): bool { |
| 31 |
return class_exists( '\\WP_HTML_Tag_Processor' ); |
| 32 |
} |
| 33 |
|
| 34 |
/** |
| 35 |
* Iterate every <link rel="stylesheet">, letting a callback rewrite the href. |
| 36 |
* |
| 37 |
* @param string $html HTML. |
| 38 |
* @param callable $callback fn(string $href, array $attrs): ?string — return new href or null to skip. |
| 39 |
* @return array{html:string,changed:bool} |
| 40 |
*/ |
| 41 |
public function rewrite_stylesheet_links( string $html, callable $callback ): array { |
| 42 |
if ( ! $this->api_ready() ) { |
| 43 |
return array( 'html' => $html, 'changed' => false ); |
| 44 |
} |
| 45 |
|
| 46 |
$processor = new WP_HTML_Tag_Processor( $html ); |
| 47 |
$changed = false; |
| 48 |
|
| 49 |
while ( $processor->next_tag( array( 'tag_name' => 'LINK' ) ) ) { |
| 50 |
$rel = (string) $processor->get_attribute( 'rel' ); |
| 51 |
|
| 52 |
if ( false === stripos( $rel, 'stylesheet' ) ) { |
| 53 |
continue; |
| 54 |
} |
| 55 |
|
| 56 |
$href = $processor->get_attribute( 'href' ); |
| 57 |
|
| 58 |
if ( ! is_string( $href ) || '' === $href ) { |
| 59 |
continue; |
| 60 |
} |
| 61 |
|
| 62 |
$attrs = array( |
| 63 |
'id' => $processor->get_attribute( 'id' ), |
| 64 |
'media' => $processor->get_attribute( 'media' ), |
| 65 |
); |
| 66 |
|
| 67 |
$new = $callback( $href, $attrs ); |
| 68 |
|
| 69 |
if ( is_string( $new ) && $new !== $href ) { |
| 70 |
$processor->set_attribute( 'href', $new ); |
| 71 |
$changed = true; |
| 72 |
} |
| 73 |
} |
| 74 |
|
| 75 |
return array( |
| 76 |
'html' => $processor->get_updated_html(), |
| 77 |
'changed' => $changed, |
| 78 |
); |
| 79 |
} |
| 80 |
|
| 81 |
/** |
| 82 |
* Collect all <link rel="stylesheet"> hrefs (read-only). |
| 83 |
* |
| 84 |
* @param string $html HTML. |
| 85 |
* @return string[] |
| 86 |
*/ |
| 87 |
public function stylesheet_hrefs( string $html ): array { |
| 88 |
if ( ! $this->api_ready() ) { |
| 89 |
return array(); |
| 90 |
} |
| 91 |
|
| 92 |
$processor = new WP_HTML_Tag_Processor( $html ); |
| 93 |
$hrefs = array(); |
| 94 |
|
| 95 |
while ( $processor->next_tag( array( 'tag_name' => 'LINK' ) ) ) { |
| 96 |
$rel = (string) $processor->get_attribute( 'rel' ); |
| 97 |
|
| 98 |
if ( false === stripos( $rel, 'stylesheet' ) ) { |
| 99 |
continue; |
| 100 |
} |
| 101 |
|
| 102 |
$href = $processor->get_attribute( 'href' ); |
| 103 |
|
| 104 |
if ( is_string( $href ) && '' !== $href ) { |
| 105 |
$hrefs[] = $href; |
| 106 |
} |
| 107 |
} |
| 108 |
|
| 109 |
return $hrefs; |
| 110 |
} |
| 111 |
|
| 112 |
/** |
| 113 |
* Physically remove specific <link> stylesheet tags by exact href. |
| 114 |
* |
| 115 |
* The tokenizer can't delete nodes, so we mark them inert with the |
| 116 |
* tokenizer (reliable matching) and then strip the marked tags by regex. |
| 117 |
* |
| 118 |
* @param string $html HTML. |
| 119 |
* @param string[] $hrefs Exact href values to remove. |
| 120 |
* @return string |
| 121 |
*/ |
| 122 |
public function remove_stylesheet_links( string $html, array $hrefs ): string { |
| 123 |
if ( empty( $hrefs ) ) { |
| 124 |
return $html; |
| 125 |
} |
| 126 |
|
| 127 |
if ( ! $this->api_ready() ) { |
| 128 |
return $html; |
| 129 |
} |
| 130 |
|
| 131 |
$targets = array_fill_keys( $hrefs, true ); |
| 132 |
$processor = new WP_HTML_Tag_Processor( $html ); |
| 133 |
$hit = false; |
| 134 |
|
| 135 |
while ( $processor->next_tag( array( 'tag_name' => 'LINK' ) ) ) { |
| 136 |
$rel = (string) $processor->get_attribute( 'rel' ); |
| 137 |
|
| 138 |
if ( false === stripos( $rel, 'stylesheet' ) ) { |
| 139 |
continue; |
| 140 |
} |
| 141 |
|
| 142 |
$href = $processor->get_attribute( 'href' ); |
| 143 |
|
| 144 |
if ( is_string( $href ) && isset( $targets[ $href ] ) ) { |
| 145 |
$processor->set_attribute( 'data-easyfonts-remove', '1' ); |
| 146 |
$hit = true; |
| 147 |
} |
| 148 |
} |
| 149 |
|
| 150 |
if ( ! $hit ) { |
| 151 |
return $html; |
| 152 |
} |
| 153 |
|
| 154 |
$updated = $processor->get_updated_html(); |
| 155 |
|
| 156 |
// Drop any <link …> carrying our marker. If PCRE bails (e.g. backtrack |
| 157 |
// limit), keep the tokenized HTML rather than blanking the document. |
| 158 |
$cleaned = preg_replace( '#<link\b[^>]*\bdata-easyfonts-remove=([\'"])1\1[^>]*/?>\s*#i', '', $updated ); |
| 159 |
|
| 160 |
return null === $cleaned ? $updated : $cleaned; |
| 161 |
} |
| 162 |
|
| 163 |
/** |
| 164 |
* Process the CSS contents of each <style> block. |
| 165 |
* |
| 166 |
* @param string $html HTML. |
| 167 |
* @param callable $callback fn(string $css): ?string — return new CSS or null to leave unchanged. |
| 168 |
* @return array{html:string,changed:bool} |
| 169 |
*/ |
| 170 |
public function rewrite_style_blocks( string $html, callable $callback ): array { |
| 171 |
$changed = false; |
| 172 |
|
| 173 |
// Linear <style>…</style> scanner — NOT a regex. A regex such as |
| 174 |
// `(<style\b[^>]*>)(.*?)(</style>)/s` catastrophically backtracks (and on |
| 175 |
// PCRE2+JIT can crash the process) over the megabyte-sized inline CSS that |
| 176 |
// page builders emit, which silently blanked the page. This strpos walk is |
| 177 |
// O(n) and cannot backtrack or overflow. |
| 178 |
$max = (int) apply_filters( 'easyfonts_max_style_block', 2 * 1024 * 1024 ); |
| 179 |
$out = ''; |
| 180 |
$offset = 0; |
| 181 |
$len = strlen( $html ); |
| 182 |
|
| 183 |
while ( $offset < $len ) { |
| 184 |
$open = stripos( $html, '<style', $offset ); |
| 185 |
|
| 186 |
if ( false === $open ) { |
| 187 |
$out .= substr( $html, $offset ); |
| 188 |
break; |
| 189 |
} |
| 190 |
|
| 191 |
// Make sure "<style" is a real tag start (next char ends the name). |
| 192 |
$boundary = $html[ $open + 6 ] ?? ''; |
| 193 |
|
| 194 |
if ( '' !== $boundary && '>' !== $boundary && '/' !== $boundary && ! ctype_space( $boundary ) ) { |
| 195 |
$out .= substr( $html, $offset, ( $open + 6 ) - $offset ); |
| 196 |
$offset = $open + 6; |
| 197 |
continue; |
| 198 |
} |
| 199 |
|
| 200 |
$gt = strpos( $html, '>', $open ); |
| 201 |
|
| 202 |
if ( false === $gt ) { |
| 203 |
$out .= substr( $html, $offset ); |
| 204 |
break; |
| 205 |
} |
| 206 |
|
| 207 |
$close = stripos( $html, '</style>', $gt + 1 ); |
| 208 |
|
| 209 |
if ( false === $close ) { |
| 210 |
$out .= substr( $html, $offset ); |
| 211 |
break; |
| 212 |
} |
| 213 |
|
| 214 |
// Copy everything up to this <style>, then the opener verbatim. |
| 215 |
$out .= substr( $html, $offset, $open - $offset ); |
| 216 |
$tag_open = substr( $html, $open, ( $gt - $open ) + 1 ); |
| 217 |
$content = substr( $html, $gt + 1, $close - ( $gt + 1 ) ); |
| 218 |
|
| 219 |
if ( strlen( $content ) <= $max ) { |
| 220 |
$new = $callback( $content ); |
| 221 |
|
| 222 |
if ( is_string( $new ) && $new !== $content ) { |
| 223 |
$changed = true; |
| 224 |
$content = $new; |
| 225 |
} |
| 226 |
} |
| 227 |
|
| 228 |
$out .= $tag_open . $content . '</style>'; |
| 229 |
$offset = $close + 8; |
| 230 |
} |
| 231 |
|
| 232 |
return array( |
| 233 |
'html' => $out, |
| 234 |
'changed' => $changed, |
| 235 |
); |
| 236 |
} |
| 237 |
|
| 238 |
/** |
| 239 |
* Remove Google-font resource hints (preconnect / dns-prefetch / preload). |
| 240 |
* |
| 241 |
* @param string $html HTML. |
| 242 |
* @param string[] $domains Domains to strip hints for. |
| 243 |
* @return array{html:string,changed:bool} |
| 244 |
*/ |
| 245 |
public function strip_font_hints( string $html, array $domains ): array { |
| 246 |
if ( ! $this->api_ready() ) { |
| 247 |
return array( 'html' => $html, 'changed' => false ); |
| 248 |
} |
| 249 |
|
| 250 |
$input = $html; |
| 251 |
$processor = new WP_HTML_Tag_Processor( $html ); |
| 252 |
$remove = array(); |
| 253 |
|
| 254 |
// Identify offending hint tags by their href (read pass). |
| 255 |
while ( $processor->next_tag( array( 'tag_name' => 'LINK' ) ) ) { |
| 256 |
$rel = strtolower( (string) $processor->get_attribute( 'rel' ) ); |
| 257 |
|
| 258 |
if ( ! in_array( $rel, array( 'preconnect', 'dns-prefetch', 'preload' ), true ) ) { |
| 259 |
continue; |
| 260 |
} |
| 261 |
|
| 262 |
$href = (string) $processor->get_attribute( 'href' ); |
| 263 |
|
| 264 |
foreach ( $domains as $domain ) { |
| 265 |
if ( false !== strpos( $href, $domain ) ) { |
| 266 |
// Neutralise: blank the href + rel so the hint is inert. |
| 267 |
$processor->set_attribute( 'href', '' ); |
| 268 |
$processor->set_attribute( 'rel', 'easyfonts-removed' ); |
| 269 |
$remove[] = true; |
| 270 |
break; |
| 271 |
} |
| 272 |
} |
| 273 |
} |
| 274 |
|
| 275 |
$html = $processor->get_updated_html(); |
| 276 |
|
| 277 |
// Physically drop the neutralised tags. If PCRE bails, fall back to the |
| 278 |
// input HTML rather than returning '' (which would blank the page). |
| 279 |
$cleaned = preg_replace( '/<link\b[^>]*\brel=([\'"])easyfonts-removed\1[^>]*>\s*/i', '', $html ); |
| 280 |
|
| 281 |
if ( null === $cleaned ) { |
| 282 |
return array( 'html' => $input, 'changed' => false ); |
| 283 |
} |
| 284 |
|
| 285 |
return array( |
| 286 |
'html' => $cleaned, |
| 287 |
'changed' => ! empty( $remove ), |
| 288 |
); |
| 289 |
} |
| 290 |
|
| 291 |
/** |
| 292 |
* Insert a snippet immediately before </head>. |
| 293 |
* |
| 294 |
* @param string $html HTML. |
| 295 |
* @param string $snippet Snippet. |
| 296 |
* @return string |
| 297 |
*/ |
| 298 |
public function before_head_close( string $html, string $snippet ): string { |
| 299 |
$pos = stripos( $html, '</head>' ); |
| 300 |
|
| 301 |
if ( false === $pos ) { |
| 302 |
return $html; |
| 303 |
} |
| 304 |
|
| 305 |
return substr_replace( $html, $snippet . "\n</head>", $pos, 7 ); |
| 306 |
} |
| 307 |
} |
| 308 |
|
