| 1 |
<?php |
| 2 |
/** |
| 3 |
* Asset minifier — HTML, CSS, JS. |
| 4 |
* |
| 5 |
* Uses matthiasmullie/minify for CSS/JS. Local enqueued assets are minified |
| 6 |
* once, cached on disk, and the loader URL is rewritten to point at the |
| 7 |
* cached file. |
| 8 |
* |
| 9 |
* @package XSpeed |
| 10 |
*/ |
| 11 |
|
| 12 |
namespace XSpeed; |
| 13 |
|
| 14 |
defined( 'ABSPATH' ) || exit; |
| 15 |
|
| 16 |
class Minifier { |
| 17 |
|
| 18 |
const MIN_SUBDIR = 'min'; |
| 19 |
|
| 20 |
/** |
| 21 |
* Absolute path to the minified-cache directory. Always derived from |
| 22 |
* XSPEED_CACHE_DIR (the plugin's own cache root) — never assembled from |
| 23 |
* arbitrary URL fragments. |
| 24 |
*/ |
| 25 |
public static function min_dir() { |
| 26 |
return trailingslashit( XSPEED_CACHE_DIR ) . self::MIN_SUBDIR; |
| 27 |
} |
| 28 |
|
| 29 |
/** |
| 30 |
* Public URL of the minified-cache directory. Built from content_url() + |
| 31 |
* the known relative path, not by string-replacing WP_CONTENT_DIR out of |
| 32 |
* a filesystem path (which would assume the filesystem layout matches |
| 33 |
* the URL layout — it does not on Bedrock-style installs, multisite with |
| 34 |
* mapped domains, or any setup with a relocated wp-content). |
| 35 |
*/ |
| 36 |
private static function min_url() { |
| 37 |
// XSPEED_CACHE_DIR lives under wp-content (defined in xspeed.php as |
| 38 |
// WP_CONTENT_DIR . '/cache/xspeed'), so the URL is content_url() + |
| 39 |
// the known suffix. We do not derive URLs from arbitrary filesystem |
| 40 |
// paths anywhere in this plugin. |
| 41 |
$url = trailingslashit( content_url( 'cache/xspeed' ) ) . self::MIN_SUBDIR; |
| 42 |
// Force the site's scheme: content_url() derives its scheme from |
| 43 |
// is_ssl(), which is false behind a TLS-terminating reverse proxy, so |
| 44 |
// it can emit an http:// URL on an https page — the browser then blocks |
| 45 |
// the minified stylesheet as mixed content and the page renders |
| 46 |
// unstyled. Match home_url()'s registered scheme instead. (FBS-83633) |
| 47 |
$scheme = wp_parse_url( home_url(), PHP_URL_SCHEME ) ?: 'https'; |
| 48 |
return set_url_scheme( $url, $scheme ); |
| 49 |
} |
| 50 |
|
| 51 |
public function __construct() { |
| 52 |
// Only run on the frontend — never minify wp-admin, AJAX, REST or cron |
| 53 |
// asset URLs. Page caching already handles the logged-in case for |
| 54 |
// the HTML response; minify scope is the public frontend. |
| 55 |
if ( is_admin() || ( defined( 'DOING_AJAX' ) && DOING_AJAX ) || ( defined( 'DOING_CRON' ) && DOING_CRON ) || ( defined( 'REST_REQUEST' ) && REST_REQUEST ) ) { |
| 56 |
return; |
| 57 |
} |
| 58 |
|
| 59 |
// A page-builder editing screen is a front-end URL, so none of the |
| 60 |
// guards above catch it. Optimizing it breaks the editor outright -- |
| 61 |
// Combine JS reorders the builder's own dependency graph and the |
| 62 |
// toolbar never renders. There is no speed to win on a logged-in, |
| 63 |
// uncacheable editing request anyway. (#281) |
| 64 |
if ( Builder_Editor::is_active() ) { |
| 65 |
return; |
| 66 |
} |
| 67 |
|
| 68 |
// Settings now live in the per-module option (xspeed_module_minify), |
| 69 |
// owned by XSpeed\Modules\Minify\MinifyModule. We read through |
| 70 |
// Settings_Manager so schema-validated values are returned even |
| 71 |
// if the option was hand-edited. |
| 72 |
$opts = Settings_Manager::get( 'minify' ); |
| 73 |
|
| 74 |
if ( ! empty( $opts['minify_css'] ) ) { |
| 75 |
add_filter( 'style_loader_src', array( __CLASS__, 'rewrite_style' ), 10, 2 ); |
| 76 |
} |
| 77 |
if ( ! empty( $opts['minify_js'] ) ) { |
| 78 |
add_filter( 'script_loader_src', array( __CLASS__, 'rewrite_script' ), 10, 2 ); |
| 79 |
// The src rewrite above runs before any plugin's own |
| 80 |
// `script_loader_tag` filter can stamp data-no-minify / |
| 81 |
// data-no-optimize onto the tag, so the marker arrives too late |
| 82 |
// to prevent it. Priority 15: after third-party tag filters at |
| 83 |
// the default 10 have printed their markers, before our defer |
| 84 |
// (20) and delay (30) look at the tag. (#456) |
| 85 |
add_filter( 'script_loader_tag', array( Minify_Filters::class, 'restore_marked_script_src' ), 15, 3 ); |
| 86 |
} |
| 87 |
|
| 88 |
// Phase 4.1a — filter-only "smarter minifier" features. Each is |
| 89 |
// gated on its own toggle so users can enable any subset. |
| 90 |
if ( ! empty( $opts['remove_query_strings'] ) ) { |
| 91 |
add_filter( 'style_loader_src', array( Minify_Filters::class, 'strip_version_query' ), 20 ); |
| 92 |
add_filter( 'script_loader_src', array( Minify_Filters::class, 'strip_version_query' ), 20 ); |
| 93 |
} |
| 94 |
if ( ! empty( $opts['defer_js'] ) ) { |
| 95 |
add_filter( 'script_loader_tag', array( Minify_Filters::class, 'defer_script_tag' ), 20, 3 ); |
| 96 |
} |
| 97 |
if ( ! empty( $opts['delay_js'] ) ) { |
| 98 |
// Delay applies a transform that's mutually exclusive with |
| 99 |
// plain defer — when both are on, delay wins (the bootstrap |
| 100 |
// will re-attach as a regular <script> on interaction). |
| 101 |
add_filter( 'script_loader_tag', array( Minify_Filters::class, 'delay_script_tag' ), 30, 3 ); |
| 102 |
add_action( 'wp_footer', array( Minify_Filters::class, 'print_delay_bootstrap' ), 1000 ); |
| 103 |
// script_loader_tag only fires for wp_enqueue_script()'d assets. |
| 104 |
// Analytics / pixel / chat-widget tags printed straight into |
| 105 |
// wp_head bypass it, and those are usually the heaviest scripts |
| 106 |
// on the page — so sweep the finished buffer too. Runs before |
| 107 |
// minify_html (same filter, default priority) and is baked into |
| 108 |
// the cache file, so it replays on static hits where PHP never |
| 109 |
// boots. |
| 110 |
add_filter( 'xspeed_cache_final_html', array( Minify_Filters::class, 'delay_raw_script_tags' ), 20 ); |
| 111 |
// The vendors' own install snippets are INLINE (no src at all), |
| 112 |
// so the src sweep above never sees them; this one parks an |
| 113 |
// inline body that names a known third-party host. |
| 114 |
add_filter( 'xspeed_cache_final_html', array( Minify_Filters::class, 'delay_inline_snippets' ), 21 ); |
| 115 |
} |
| 116 |
if ( ! empty( $opts['async_css'] ) ) { |
| 117 |
add_filter( 'style_loader_tag', array( Minify_Filters::class, 'async_style_tag' ), 20, 2 ); |
| 118 |
// style_loader_tag only fires for wp_enqueue_style()'d sheets. |
| 119 |
// Themes print Google/Bunny/Typekit font CSS as literal <link> |
| 120 |
// markup in the head, so those sheets never reach the filter and |
| 121 |
// stay render-blocking — sweep the finished buffer for the known |
| 122 |
// font-CSS hosts. Baked into the cache file, so it replays on |
| 123 |
// static hits where PHP never boots. |
| 124 |
add_filter( 'xspeed_cache_final_html', array( Minify_Filters::class, 'async_raw_font_css_links' ), 22 ); |
| 125 |
} |
| 126 |
|
| 127 |
/* |
| 128 |
* CSS combining runs on the FINISHED HTML, not the enqueue queue. |
| 129 |
* |
| 130 |
* The queue-walking version could not be made correct: whatever it |
| 131 |
* wrote at priority 999, WordPress edited afterwards. Core's |
| 132 |
* wp_maybe_inline_styles() inlines any queued handle carrying a `path` |
| 133 |
* and sets src=false on it, which silently threw away the combined URL |
| 134 |
* and took the sheets we had blanked with it — six stylesheets became |
| 135 |
* one and the site rendered unstyled. See Css_Combine_Buffer's header |
| 136 |
* for the full trace. (#195) |
| 137 |
* |
| 138 |
* Two entry points, because the page cache's filter is not always |
| 139 |
* available: `xspeed_cache_final_html` fires only on a cacheable MISS, |
| 140 |
* so on a site with the cache off — or on an excluded URL like /cart — |
| 141 |
* combining would silently stop working. Css_Combine_Buffer::boot() |
| 142 |
* opens its own buffer in exactly those cases and no-ops otherwise, so |
| 143 |
* the page is transformed once either way. |
| 144 |
*/ |
| 145 |
if ( ! empty( $opts['combine_css'] ) ) { |
| 146 |
add_filter( 'xspeed_cache_final_html', array( Css_Combine_Buffer::class, 'process' ), 5 ); |
| 147 |
Css_Combine_Buffer::boot(); |
| 148 |
} |
| 149 |
if ( ! empty( $opts['combine_js'] ) ) { |
| 150 |
// JS stays on the enqueue path for now: dependency order, |
| 151 |
// async/defer and wp_add_inline_script make it a different |
| 152 |
// problem, and the reported break is CSS-only. Moving it is worth |
| 153 |
// its own change rather than doubling the blast radius here. |
| 154 |
add_action( 'wp_enqueue_scripts', array( Asset_Combiner::class, 'combine_scripts' ), 999 ); |
| 155 |
} |
| 156 |
} |
| 157 |
|
| 158 |
/** |
| 159 |
* HTML elements that participate in an inline formatting context, where |
| 160 |
* whitespace between two of them renders as a visible space. |
| 161 |
* |
| 162 |
* Deliberately excludes <br> (nothing to separate) and replaced/embedded |
| 163 |
* inline elements that sit alone. Anything not listed is treated as block |
| 164 |
* level, where inter-tag whitespace collapses to nothing and is safe to |
| 165 |
* strip. (FBS-84090) |
| 166 |
*/ |
| 167 |
private const INLINE_TAGS = array( |
| 168 |
'a', 'abbr', 'b', 'bdi', 'bdo', 'cite', 'code', 'data', 'del', 'dfn', |
| 169 |
'em', 'i', 'ins', 'kbd', 'label', 'mark', 'q', 'rp', 'rt', 'ruby', |
| 170 |
's', 'samp', 'small', 'span', 'strong', 'sub', 'sup', 'time', 'u', |
| 171 |
'var', 'wbr', 'img', 'button', 'select', 'output', |
| 172 |
); |
| 173 |
|
| 174 |
/** True when $tag renders inline, so whitespace beside it is visible. */ |
| 175 |
private static function is_inline( string $tag ): bool { |
| 176 |
return in_array( strtolower( $tag ), self::INLINE_TAGS, true ); |
| 177 |
} |
| 178 |
|
| 179 |
/** |
| 180 |
* Why minification is being skipped, when it is. '' when it will run. |
| 181 |
* |
| 182 |
* minify_html can read "on" in every settings surface while producing |
| 183 |
* byte-identical HTML, because the guard below silently returns the |
| 184 |
* input. Field report: a live site showed `minify_html: on` with 3,856 |
| 185 |
* indented lines in the delivered HTML and nothing anywhere explaining |
| 186 |
* the contradiction — the setting looked broken rather than suppressed. |
| 187 |
* Callers that report status MUST consult this so the refusal is |
| 188 |
* visible. (Same class as Cache::static_rewrite_block_reason().) |
| 189 |
* |
| 190 |
* @return string 'wp_debug', 'filter', or ''. |
| 191 |
*/ |
| 192 |
public static function skip_reason(): string { |
| 193 |
$debug_skip = defined( 'WP_DEBUG' ) && WP_DEBUG; |
| 194 |
if ( ! apply_filters( 'xspeed_skip_minify', $debug_skip ) ) { |
| 195 |
return ''; |
| 196 |
} |
| 197 |
// Distinguish the built-in WP_DEBUG rule from a third party |
| 198 |
// filtering the escape hatch — the fixes are different. |
| 199 |
return $debug_skip ? 'wp_debug' : 'filter'; |
| 200 |
} |
| 201 |
|
| 202 |
public static function minify_html( $html ) { |
| 203 |
if ( '' !== self::skip_reason() ) { |
| 204 |
return $html; |
| 205 |
} |
| 206 |
|
| 207 |
$placeholders = array(); |
| 208 |
$pattern = '#<(pre|textarea|script|style)\b[^>]*>.*?</\1>#is'; |
| 209 |
$html = preg_replace_callback( |
| 210 |
$pattern, |
| 211 |
function ( $m ) use ( &$placeholders ) { |
| 212 |
$key = '__XSPEED_PH_' . count( $placeholders ) . '__'; |
| 213 |
// The placeholder pass exists to protect content whose |
| 214 |
// whitespace is significant (<pre>, <textarea>) and to keep |
| 215 |
// the tag-boundary regex off script bodies. <style> and |
| 216 |
// <script> were grouped in with them, so protection became a |
| 217 |
// permanent exemption: on builder sites where most CSS is |
| 218 |
// inline, a page with minify ON shipped fully indented. Minify |
| 219 |
// the BODY here, before it's stashed, so the outer passes |
| 220 |
// still never see it. (#2) |
| 221 |
$placeholders[ $key ] = self::minify_inline_block( $m[0], strtolower( $m[1] ) ); |
| 222 |
return $key; |
| 223 |
}, |
| 224 |
$html |
| 225 |
); |
| 226 |
|
| 227 |
$html = preg_replace( '/<!--(?!\[if).*?-->/s', '', $html ); |
| 228 |
$html = preg_replace( '/\s+/', ' ', $html ); |
| 229 |
|
| 230 |
/* |
| 231 |
* Collapse whitespace BETWEEN TAGS — but never where it is visible. |
| 232 |
* |
| 233 |
* Whitespace separating two INLINE elements is a real, rendered space: |
| 234 |
* WooCommerce emits `</del> <ins>` for a sale price, and that single |
| 235 |
* character is the gap between "$32.50" and "$29.50". Stripping it |
| 236 |
* printed "$32.50$29.50" run together, and only with cache on — the |
| 237 |
* un-minified page was fine. (FBS-84090) |
| 238 |
* |
| 239 |
* So the strip only applies when at least one side is a BLOCK-level |
| 240 |
* (or non-rendered) tag, where the whitespace collapses away anyway. |
| 241 |
* Inline-to-inline boundaries keep their single space. |
| 242 |
*/ |
| 243 |
$html = preg_replace_callback( |
| 244 |
// left tag name (may be a closing tag) … whitespace … right tag name |
| 245 |
'#</?([a-zA-Z][a-zA-Z0-9-]*)\b[^>]*>\s+<(/?)([a-zA-Z][a-zA-Z0-9-]*)#', |
| 246 |
static function ( $m ) { |
| 247 |
// Keep the space only when BOTH sides are inline elements — |
| 248 |
// that is the one case where it is actually rendered. |
| 249 |
$keep = self::is_inline( $m[1] ) && self::is_inline( $m[3] ); |
| 250 |
$open = substr( $m[0], 0, strrpos( $m[0], '<' ) ); // through the left tag's '>' |
| 251 |
return rtrim( $open ) . ( $keep ? ' ' : '' ) . '<' . $m[2] . $m[3]; |
| 252 |
}, |
| 253 |
$html |
| 254 |
); |
| 255 |
$html = trim( $html ); |
| 256 |
|
| 257 |
foreach ( $placeholders as $key => $original ) { |
| 258 |
$html = str_replace( $key, $original, $html ); |
| 259 |
} |
| 260 |
|
| 261 |
return $html; |
| 262 |
} |
| 263 |
|
| 264 |
public static function rewrite_style( $src, $handle ) { |
| 265 |
unset( $handle ); |
| 266 |
return self::rewrite_asset( $src, 'css' ); |
| 267 |
} |
| 268 |
|
| 269 |
public static function rewrite_script( $src, $handle ) { |
| 270 |
$rewritten = self::rewrite_asset( $src, 'js' ); |
| 271 |
|
| 272 |
// Remember the pre-minify URL for this handle. script_loader_tag |
| 273 |
// runs later and only ever sees the rewritten src (a hashed |
| 274 |
// /cache/xspeed/min/<key>.js path), so a user's URL-substring |
| 275 |
// delay/exclusion target would never match once minification is |
| 276 |
// on. Minify_Filters::original_src() gives those checks the URL |
| 277 |
// the user actually wrote their target against. (FBS field report) |
| 278 |
if ( is_string( $handle ) && '' !== $handle && is_string( $src ) && $src !== $rewritten ) { |
| 279 |
Minify_Filters::remember_original_src( $handle, $src ); |
| 280 |
} |
| 281 |
|
| 282 |
return $rewritten; |
| 283 |
} |
| 284 |
|
| 285 |
/** |
| 286 |
* Replace a local CSS/JS URL with a cached, minified equivalent. |
| 287 |
* |
| 288 |
* @param string $src Original asset URL. |
| 289 |
* @param string $type 'css' or 'js'. |
| 290 |
* @return string Possibly rewritten URL. |
| 291 |
*/ |
| 292 |
private static function rewrite_asset( $src, $type ) { |
| 293 |
if ( ! is_string( $src ) || '' === $src ) { |
| 294 |
return $src; |
| 295 |
} |
| 296 |
|
| 297 |
// Skip already-minified files. |
| 298 |
if ( false !== strpos( $src, '.min.' ) ) { |
| 299 |
return $src; |
| 300 |
} |
| 301 |
|
| 302 |
// Skip anything we already produced. The Asset_Combiner minifies the |
| 303 |
// combined body itself before writing combined-<hash>.css under |
| 304 |
// min/combined/ (issue #331 — that used to be asserted here but was |
| 305 |
// not actually true, so the artifact shipped unminified), and enqueues it |
| 306 |
// as `xspeed-combined-css`; the per-file minifier used to re-minify |
| 307 |
// that combined output into a SECOND file (min/<hash2>.css) with its |
| 308 |
// own mtime-derived hash. The served HTML then pinned that second |
| 309 |
// hash, so a purge/regeneration (which changes the combined file's |
| 310 |
// mtime -> a new hash2) left the cached page pointing at a file that |
| 311 |
// no longer existed -> 404 -> unstyled/broken frontend. Leaving our |
| 312 |
// own cache output untouched keeps a single, stable URL end-to-end. |
| 313 |
if ( false !== strpos( $src, '/cache/xspeed/' ) ) { |
| 314 |
return $src; |
| 315 |
} |
| 316 |
|
| 317 |
// Resolve to a local path; bail if external or unresolvable. |
| 318 |
$path = self::url_to_path( $src ); |
| 319 |
if ( ! $path || ! is_readable( $path ) ) { |
| 320 |
return $src; |
| 321 |
} |
| 322 |
|
| 323 |
// Build a cache filename keyed on path + mtime so edits invalidate. |
| 324 |
$mtime = filemtime( $path ); |
| 325 |
$key = md5( $path . '|' . $mtime ); |
| 326 |
$cache = self::cache_path( $key, $type ); |
| 327 |
|
| 328 |
if ( ! file_exists( $cache ) ) { |
| 329 |
$ok = self::minify_file( $path, $cache, $type ); |
| 330 |
if ( ! $ok ) { |
| 331 |
return $src; |
| 332 |
} |
| 333 |
} |
| 334 |
|
| 335 |
// Return a URL to the cached file. Built from known constants — never |
| 336 |
// from str_replace on a filesystem path (which would assume the FS |
| 337 |
// layout mirrors the URL layout). |
| 338 |
return self::min_url() . '/' . $key . '.' . $type; |
| 339 |
} |
| 340 |
|
| 341 |
private static function minify_file( $source_path, $target_path, $type ) { |
| 342 |
if ( ! class_exists( '\\MatthiasMullie\\Minify\\CSS' ) ) { |
| 343 |
return false; |
| 344 |
} |
| 345 |
|
| 346 |
// Path-traversal guard: refuse to write anywhere outside our cache |
| 347 |
// dir, even if a malicious filter ever produced a poisoned key. |
| 348 |
$cache_root = self::min_dir(); |
| 349 |
self::ensure_dir( $cache_root ); |
| 350 |
$real_root = realpath( $cache_root ); |
| 351 |
$real_dir = realpath( dirname( $target_path ) ); |
| 352 |
if ( ! $real_root || ! $real_dir || 0 !== strpos( $real_dir, $real_root ) ) { |
| 353 |
return false; |
| 354 |
} |
| 355 |
|
| 356 |
try { |
| 357 |
if ( 'css' === $type ) { |
| 358 |
// Passing the TARGET path makes matthiasmullie/minify rebase every |
| 359 |
// relative url(...) / @import against the minified file's location. |
| 360 |
// Without it, a stylesheet moved from e.g. |
| 361 |
// .../font-awesome/css/all.css to cache/xspeed/min/<key>.css keeps |
| 362 |
// its original url(../webfonts/…) — which then resolves against the |
| 363 |
// cache dir and 404s (missing FontAwesome/eicons/WooCommerce fonts). |
| 364 |
$minifier = new \MatthiasMullie\Minify\CSS( $source_path ); |
| 365 |
$minified = $minifier->minify( $target_path ); |
| 366 |
return '' !== $minified && file_exists( $target_path ); |
| 367 |
} |
| 368 |
|
| 369 |
$minifier = new \MatthiasMullie\Minify\JS( $source_path ); |
| 370 |
$minified = $minifier->minify(); |
| 371 |
|
| 372 |
// Sanity check: paren/brace/bracket/backtick balance must be preserved. |
| 373 |
// matthiasmullie/minify can silently truncate mid-template-literal on |
| 374 |
// complex modern JS — bail rather than ship a broken file. |
| 375 |
// phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents -- WP_Filesystem requires admin context; minification runs on frontend page renders. Source already validated as readable on line 121. |
| 376 |
$source = file_get_contents( $source_path ); |
| 377 |
if ( false === $source || ! self::balanced( $source, $minified ) ) { |
| 378 |
return false; |
| 379 |
} |
| 380 |
|
| 381 |
// phpcs:ignore WordPress.WP.AlternativeFunctions.file_put_contents_file_put_contents -- WP_Filesystem requires admin context; minification runs on frontend page renders. |
| 382 |
$bytes = file_put_contents( $target_path, $minified ); |
| 383 |
return false !== $bytes && file_exists( $target_path ); |
| 384 |
} catch ( \Throwable $e ) { |
| 385 |
return false; |
| 386 |
} |
| 387 |
} |
| 388 |
|
| 389 |
/** |
| 390 |
* Cheap structural sanity check between source + minified bodies. |
| 391 |
* |
| 392 |
* Counts paired-delimiter tokens (parens, braces, brackets, backticks) |
| 393 |
* in each and bails when the counts disagree — matthiasmullie/minify |
| 394 |
* has been observed to silently truncate inside template literals on |
| 395 |
* complex modern JS (see commit history), shipping a body that LOOKS |
| 396 |
* minified but is structurally broken and crashes the page at parse. |
| 397 |
* |
| 398 |
* Backticks are paired (open + close = same token), so the count |
| 399 |
* itself must match exactly. Strings inside the source can contain |
| 400 |
* literal `{` / `}` / `[` / `]` that throw off the count by the same |
| 401 |
* amount in both bodies (since they survive minification as-is), so |
| 402 |
* the equality check is robust to that noise. |
| 403 |
*/ |
| 404 |
/** |
| 405 |
* Minify the body of one captured inline block, or return it untouched. |
| 406 |
* |
| 407 |
* Only `<style>` and JavaScript `<script>` bodies are eligible: |
| 408 |
* |
| 409 |
* - `<pre>` / `<textarea>` — whitespace is rendered, never touch it. |
| 410 |
* - `<script>` with a non-JS `type` — `application/ld+json`, |
| 411 |
* `text/template`, `text/x-handlebars` and anything unrecognised are |
| 412 |
* data or markup, not code. Minifying JSON-LD would corrupt structured |
| 413 |
* data; minifying a template would eat the markup it holds. An unknown |
| 414 |
* type is treated as non-JS on purpose: guessing wrong breaks the page, |
| 415 |
* while skipping only forgoes a few bytes. |
| 416 |
* - `<script src="...">` — the body is empty; the file path already goes |
| 417 |
* through minify_file(). |
| 418 |
* |
| 419 |
* Every result is checked with balanced(), the same structural guard the |
| 420 |
* file path uses, so a body the library truncates is shipped as-is rather |
| 421 |
* than broken. (#2) |
| 422 |
* |
| 423 |
* @param string $block Full matched tag, opening tag through closing tag. |
| 424 |
* @param string $tag Lowercased tag name. |
| 425 |
* @return string Minified block, or $block unchanged. |
| 426 |
*/ |
| 427 |
private static function minify_inline_block( string $block, string $tag ): string { |
| 428 |
if ( 'style' !== $tag && 'script' !== $tag ) { |
| 429 |
return $block; // pre / textarea — significant whitespace. |
| 430 |
} |
| 431 |
if ( ! class_exists( '\\MatthiasMullie\\Minify\\CSS' ) ) { |
| 432 |
return $block; |
| 433 |
} |
| 434 |
|
| 435 |
// Split into opening tag / body / closing tag. Anything that doesn't |
| 436 |
// match this shape isn't something we should be rewriting. |
| 437 |
if ( ! preg_match( '#^(<' . $tag . '\b[^>]*>)(.*)(</' . $tag . '\s*>)$#is', $block, $parts ) ) { |
| 438 |
return $block; |
| 439 |
} |
| 440 |
list( , $open, $body, $close ) = $parts; |
| 441 |
|
| 442 |
if ( '' === trim( $body ) ) { |
| 443 |
return $block; |
| 444 |
} |
| 445 |
|
| 446 |
// The tag asked to be left alone (data-no-optimize / data-no-minify — |
| 447 |
// the convention consent managers print on their config scripts). (#456) |
| 448 |
if ( Minify_Filters::tag_opts_out( $open ) ) { |
| 449 |
return $block; |
| 450 |
} |
| 451 |
|
| 452 |
// Refuse a body that is already structurally broken. balanced() only |
| 453 |
// compares source against minified, so it passes when BOTH are equally |
| 454 |
// unbalanced — `function x( {` minifies to `function x({`, same counts, |
| 455 |
// guard satisfied, broken code reformatted. Rewriting a body we can't |
| 456 |
// parse risks turning a page that happens to work into one that does |
| 457 |
// not, for no gain. (#2 AC: a syntactically broken block is left |
| 458 |
// untouched.) |
| 459 |
if ( ! self::self_consistent( $body ) ) { |
| 460 |
return $block; |
| 461 |
} |
| 462 |
|
| 463 |
if ( 'script' === $tag ) { |
| 464 |
// An external script has no body worth minifying. |
| 465 |
if ( preg_match( '#\bsrc\s*=#i', $open ) ) { |
| 466 |
return $block; |
| 467 |
} |
| 468 |
// No type, or an explicitly JavaScript type, is code. Everything |
| 469 |
// else is data/markup — see the docblock. |
| 470 |
$js_types = array( |
| 471 |
'text/javascript', |
| 472 |
'application/javascript', |
| 473 |
'application/ecmascript', |
| 474 |
'text/ecmascript', |
| 475 |
'module', |
| 476 |
); |
| 477 |
if ( preg_match( '#\btype\s*=\s*["\']?([^"\'\s>]+)#i', $open, $type_match ) ) { |
| 478 |
if ( ! in_array( strtolower( trim( $type_match[1] ) ), $js_types, true ) ) { |
| 479 |
return $block; |
| 480 |
} |
| 481 |
} |
| 482 |
} |
| 483 |
|
| 484 |
try { |
| 485 |
$minifier = 'style' === $tag |
| 486 |
? new \MatthiasMullie\Minify\CSS() |
| 487 |
: new \MatthiasMullie\Minify\JS(); |
| 488 |
$minifier->add( $body ); |
| 489 |
$minified = $minifier->minify(); |
| 490 |
} catch ( \Throwable $e ) { |
| 491 |
return $block; |
| 492 |
} |
| 493 |
|
| 494 |
// A minifier that returns nothing for a non-empty body has failed, not |
| 495 |
// succeeded — shipping '' would silently delete the rule set. |
| 496 |
if ( ! is_string( $minified ) || '' === trim( $minified ) ) { |
| 497 |
return $block; |
| 498 |
} |
| 499 |
if ( ! self::balanced( $body, $minified ) ) { |
| 500 |
return $block; |
| 501 |
} |
| 502 |
|
| 503 |
return $open . $minified . $close; |
| 504 |
} |
| 505 |
|
| 506 |
/** |
| 507 |
* Does a body's own paired delimiters balance? |
| 508 |
* |
| 509 |
* balanced() is a RELATIVE check — source against minified — so it cannot |
| 510 |
* see input that was already broken: an unbalanced body minifies to an |
| 511 |
* equally unbalanced one and the counts still agree. This is the absolute |
| 512 |
* check, applied to the source alone before we touch it. |
| 513 |
* |
| 514 |
* Deliberately naive: it counts tokens without parsing, so a brace inside |
| 515 |
* a string or comment skews it. That only ever makes it MORE conservative — |
| 516 |
* a false negative skips minification, which costs bytes, while a false |
| 517 |
* positive would ship broken code. (#2) |
| 518 |
* |
| 519 |
* @param string $body Inline block body. |
| 520 |
*/ |
| 521 |
private static function self_consistent( string $body ): bool { |
| 522 |
$pairs = array( |
| 523 |
'{' => '}', |
| 524 |
'(' => ')', |
| 525 |
'[' => ']', |
| 526 |
); |
| 527 |
foreach ( $pairs as $open => $close ) { |
| 528 |
if ( substr_count( $body, $open ) !== substr_count( $body, $close ) ) { |
| 529 |
return false; |
| 530 |
} |
| 531 |
} |
| 532 |
// Backticks and quotes pair with themselves, so an odd count means an |
| 533 |
// unterminated literal. |
| 534 |
foreach ( array( '`' ) as $token ) { |
| 535 |
if ( 0 !== substr_count( $body, $token ) % 2 ) { |
| 536 |
return false; |
| 537 |
} |
| 538 |
} |
| 539 |
return true; |
| 540 |
} |
| 541 |
|
| 542 |
private static function balanced( string $source, string $minified ): bool { |
| 543 |
unset( $source ); |
| 544 |
|
| 545 |
// Judge the OUTPUT, not the difference between input and output. |
| 546 |
// |
| 547 |
// This used to compare token counts across the pair, on the stated |
| 548 |
// assumption that "literal braces inside strings survive minification |
| 549 |
// unchanged, so they cancel out". Comments do not: stripping them is |
| 550 |
// the minifier's whole job, and every brace, bracket and backtick |
| 551 |
// inside one disappears with it. So any file whose comments contain a |
| 552 |
// delimiter — a commented-out block, a URL in a docblock, an SVG in a |
| 553 |
// note — failed the check and silently shipped unminified. |
| 554 |
// |
| 555 |
// It is not a rare shape. EmbedPress's front.js counts 372 braces |
| 556 |
// against 368, 61 brackets against 58 and 110 backticks against 102 |
| 557 |
// purely from comment removal, so 67 KB shipped raw where 46 KB was |
| 558 |
// correct — and `node --check` confirms that rejected output parses |
| 559 |
// fine. A guard that refuses valid work is not conservative, it is |
| 560 |
// broken: it costs bytes on every request and reports nothing. |
| 561 |
// |
| 562 |
// What the guard is FOR still stands (#2): matthiasmullie/minify can |
| 563 |
// truncate inside a template literal on complex modern JS and return a |
| 564 |
// body that looks minified but is structurally broken. That failure is |
| 565 |
// visible in the output alone — an unterminated literal leaves an odd |
| 566 |
// backtick count and unmatched braces — which is exactly what |
| 567 |
// self_consistent() measures, without the false positives. |
| 568 |
return self::self_consistent( $minified ); |
| 569 |
} |
| 570 |
|
| 571 |
/** |
| 572 |
* Resolve a local asset URL to a filesystem path using a strict allowlist |
| 573 |
* of "URL prefix → filesystem prefix" pairs registered with WordPress. |
| 574 |
* |
| 575 |
* We never assume `site_url()` maps to `ABSPATH` (the WordPress root can |
| 576 |
* live above the document root in Bedrock-style installs, behind a proxy, |
| 577 |
* or on multisite with mapped domains). Each branch resolves through a |
| 578 |
* known WP API (plugins, themes, content, includes) and validates that |
| 579 |
* `realpath()` of the result still lives under the expected base — so a |
| 580 |
* crafted `..`-laden URL cannot escape into the filesystem. |
| 581 |
* |
| 582 |
* @param string $url Asset URL (may be protocol-relative or absolute). |
| 583 |
* @return string|false Absolute filesystem path on success, false otherwise. |
| 584 |
*/ |
| 585 |
private static function url_to_path( $url ) { |
| 586 |
if ( ! is_string( $url ) || '' === $url ) { |
| 587 |
return false; |
| 588 |
} |
| 589 |
|
| 590 |
// Drop query string + fragment. |
| 591 |
$clean = strtok( $url, '?#' ); |
| 592 |
|
| 593 |
// Normalise protocol-relative + scheme variants of the host so we |
| 594 |
// match regardless of whether the asset URL came in over http/https. |
| 595 |
$site_host = wp_parse_url( home_url(), PHP_URL_HOST ); |
| 596 |
if ( 0 === strpos( $clean, '//' ) ) { |
| 597 |
$clean = 'https:' . $clean; |
| 598 |
} |
| 599 |
if ( $site_host ) { |
| 600 |
$asset_host = wp_parse_url( $clean, PHP_URL_HOST ); |
| 601 |
if ( $asset_host && $asset_host !== $site_host ) { |
| 602 |
return false; // External asset — never touch. |
| 603 |
} |
| 604 |
} |
| 605 |
|
| 606 |
$candidates = array( |
| 607 |
array( plugins_url(), WP_PLUGIN_DIR ), |
| 608 |
array( get_stylesheet_directory_uri(), get_stylesheet_directory() ), |
| 609 |
array( get_template_directory_uri(), get_template_directory() ), |
| 610 |
array( content_url(), WP_CONTENT_DIR ), |
| 611 |
array( includes_url(), ABSPATH . WPINC ), |
| 612 |
); |
| 613 |
|
| 614 |
// The host check above normalised the HOST but not the SCHEME, and the |
| 615 |
// prefix match below is a plain string compare — so an https asset URL |
| 616 |
// never matched an http base and the file silently shipped unminified. |
| 617 |
// That is not a corner case: WP_CONTENT_URL is derived from a stored |
| 618 |
// option, `plugins_url()` from another, and a site moved to https |
| 619 |
// without rewriting every row (or one behind a TLS-terminating proxy |
| 620 |
// where `is_ssl()` reads false) serves https pages off http-rooted |
| 621 |
// bases all day. Comparing scheme-less is the whole fix; the host |
| 622 |
// equality test already did the security work of refusing anything |
| 623 |
// off-site, and this runs after it. |
| 624 |
$strip_scheme = static function ( string $value ): string { |
| 625 |
return (string) preg_replace( '#^https?://#i', '//', $value ); |
| 626 |
}; |
| 627 |
$clean_match = $strip_scheme( $clean ); |
| 628 |
|
| 629 |
foreach ( $candidates as $pair ) { |
| 630 |
list( $url_base, $path_base ) = $pair; |
| 631 |
if ( ! $url_base || ! $path_base ) { |
| 632 |
continue; |
| 633 |
} |
| 634 |
$url_base = rtrim( $url_base, '/' ); |
| 635 |
$base_match = $strip_scheme( $url_base ); |
| 636 |
if ( 0 !== strpos( $clean_match, $base_match . '/' ) && $clean_match !== $base_match ) { |
| 637 |
continue; |
| 638 |
} |
| 639 |
|
| 640 |
// Slice the scheme-less pair, not the original. `https://…` and |
| 641 |
// `http://…` differ by one byte, so an offset taken from the base |
| 642 |
// as written would cut one character short of (or past) the path |
| 643 |
// when the two schemes disagree — which is the case this fix |
| 644 |
// exists for. |
| 645 |
$relative = ltrim( substr( $clean_match, strlen( $base_match ) ), '/' ); |
| 646 |
$candidate = trailingslashit( $path_base ) . $relative; |
| 647 |
|
| 648 |
$real_base = realpath( $path_base ); |
| 649 |
$real = realpath( $candidate ); |
| 650 |
if ( ! $real_base || ! $real ) { |
| 651 |
return false; |
| 652 |
} |
| 653 |
// Guard against `..`-traversal: resolved path must stay inside |
| 654 |
// the registered base. |
| 655 |
if ( 0 !== strpos( $real, $real_base ) ) { |
| 656 |
return false; |
| 657 |
} |
| 658 |
return $real; |
| 659 |
} |
| 660 |
|
| 661 |
return false; |
| 662 |
} |
| 663 |
|
| 664 |
private static function cache_path( $key, $type ) { |
| 665 |
return self::min_dir() . '/' . $key . '.' . $type; |
| 666 |
} |
| 667 |
|
| 668 |
private static function ensure_dir( $dir ) { |
| 669 |
if ( ! file_exists( $dir ) ) { |
| 670 |
wp_mkdir_p( $dir ); |
| 671 |
Cache::write_silence( $dir ); |
| 672 |
} |
| 673 |
} |
| 674 |
|
| 675 |
/** |
| 676 |
* Clear every minified / combined asset. |
| 677 |
* |
| 678 |
* @return int Files removed. Most callers are `add_action` callbacks and |
| 679 |
* ignore it; `wp xspeed purge` reports it as a line item. |
| 680 |
*/ |
| 681 |
public static function purge_minified() { |
| 682 |
return self::rmtree_files( self::min_dir() ); |
| 683 |
} |
| 684 |
|
| 685 |
/** |
| 686 |
* Recursively delete every file under $dir (and the emptied |
| 687 |
* subdirectories), keeping $dir itself. The previous glob('$dir/*') |
| 688 |
* was non-recursive and no-ops on directories, so combined assets in |
| 689 |
* min/combined/ were never cleared — a purge left a stale |
| 690 |
* combined-<hash>.css the regenerated page no longer referenced. |
| 691 |
* (FBS-83114 / FBS-83116) |
| 692 |
*/ |
| 693 |
private static function rmtree_files( string $dir ): int { |
| 694 |
if ( ! is_dir( $dir ) ) { |
| 695 |
return 0; |
| 696 |
} |
| 697 |
$removed = 0; |
| 698 |
foreach ( (array) glob( $dir . '/*' ) as $path ) { |
| 699 |
if ( is_dir( $path ) ) { |
| 700 |
$removed += self::rmtree_files( $path ); |
| 701 |
@rmdir( $path ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_rmdir, WordPress.PHP.NoSilencedErrors.Discouraged -- best-effort cleanup of our own cache subdir; WP_Filesystem is unavailable on the frontend purge path. |
| 702 |
continue; |
| 703 |
} |
| 704 |
wp_delete_file( $path ); |
| 705 |
++$removed; |
| 706 |
} |
| 707 |
return $removed; |
| 708 |
} |
| 709 |
} |
| 710 |
|
