| @@ -82,8 +82,385 @@ | ||
| 82 | 82 | self::$original_src = array(); |
| 83 | 83 | } |
| 84 | 84 | |
| 85 | 85 | /** |
| 86 | + * Does this tag (or attribute string) opt out of optimization? | |
| 87 | + * | |
| 88 | + * `data-no-optimize` / `data-no-minify` are the de-facto convention | |
| 89 | + * consent managers and other plugins print so optimizers keep hands | |
| 90 | + * off (Borlabs Cookie stamps both on its config script). The CSS | |
| 91 | + * combine buffer has honored `data-no-optimize` from the start; the | |
| 92 | + * JS paths did not, so a marked consent script was still minified | |
| 93 | + * into a hashed cache file — and a stale copy of a legally relevant | |
| 94 | + * consent config is a correctness problem, not a cosmetic one. (#456) | |
| 95 | + * | |
| 96 | + * @param string $tag A full tag, or just its attribute string. | |
| 97 | + */ | |
| 98 | + public static function tag_opts_out( string $tag ): bool { | |
| 99 | + return (bool) preg_match( '#\sdata-no-(?:optimize|minify)\b#i', $tag ); | |
| 100 | + } | |
| 101 | + | |
| 102 | + /** | |
| 103 | + * Pristine tags as they looked before any of our transforms, keyed by | |
| 104 | + * handle. See snapshot_tag() / revert_late_marked_tag(). | |
| 105 | + * | |
| 106 | + * @var array<string,string> | |
| 107 | + */ | |
| 108 | + private static $pristine_tag = array(); | |
| 109 | + | |
| 110 | + /** | |
| 111 | + * Priority for the late opt-out re-check. Past Borlabs' ScriptBlocker | |
| 112 | + * at 999 — the highest stamper we have seen in the wild — so the | |
| 113 | + * marker has certainly landed by the time we look. (#469) | |
| 114 | + */ | |
| 115 | + private const LATE_OPT_OUT_PRIORITY = 1000; | |
| 116 | + | |
| 117 | + /** | |
| 118 | + * The priority the late opt-out re-check runs at. | |
| 119 | + * | |
| 120 | + * A site whose stamper hooks even later can move ours past it. | |
| 121 | + */ | |
| 122 | + public static function late_opt_out_priority(): int { | |
| 123 | + /** | |
| 124 | + * Filter the priority of xSpeed's late data-no-optimize re-check. | |
| 125 | + * | |
| 126 | + * @param int $priority Default 1000. | |
| 127 | + */ | |
| 128 | + return (int) apply_filters( 'xspeed_late_opt_out_priority', self::LATE_OPT_OUT_PRIORITY ); | |
| 129 | + } | |
| 130 | + | |
| 131 | + /** | |
| 132 | + * Filter: `script_loader_tag`, priority 9 — remember the tag before we | |
| 133 | + * touch it, so a marker stamped later can still be honored. | |
| 134 | + * | |
| 135 | + * Our three opt-out-aware transforms run at 15/20/30. A plugin that | |
| 136 | + * stamps `data-no-optimize` AFTER them is invisible to all three: | |
| 137 | + * Borlabs Cookie stamps at priority 100, so its consent config was | |
| 138 | + * still minified into a hashed cache file AND delayed — the script | |
| 139 | + * that has to run before anything else on the page ran only on first | |
| 140 | + * interaction. Snapshotting here is what lets the late pass put the | |
| 141 | + * original back verbatim, rather than trying to unpick each transform | |
| 142 | + * in reverse. (#469) | |
| 143 | + * | |
| 144 | + * @param string $tag | |
| 145 | + * @param string $handle | |
| 146 | + * @param string $src | |
| 147 | + */ | |
| 148 | + public static function snapshot_tag( $tag, $handle, $src ): string { | |
| 149 | + if ( is_string( $tag ) && '' !== $tag && '' !== (string) $handle ) { | |
| 150 | + self::$pristine_tag[ (string) $handle ] = $tag; | |
| 151 | + } | |
| 152 | + return (string) $tag; | |
| 153 | + } | |
| 154 | + | |
| 155 | + /** | |
| 156 | + * Filter: `script_loader_tag`, priority `LATE_OPT_OUT_PRIORITY` — hand | |
| 157 | + * back the untouched tag when a late filter stamped an opt-out marker | |
| 158 | + * after our transforms had already run. | |
| 159 | + * | |
| 160 | + * The priority has to clear the stamper, not merely the transforms: | |
| 161 | + * Borlabs stamps at 100 and Borlabs' own script blocker at 999, so an | |
| 162 | + * earlier hook reads a tag whose marker has not landed yet. PHP_INT_MAX | |
| 163 | + * would be unfriendly to a site that legitimately wants the last word, | |
| 164 | + * so this sits just past the highest stamper we know of and is | |
| 165 | + * filterable. Reverting to the snapshot is deliberate: undoing | |
| 166 | + * a delay rewrite in place would mean re-deriving `src` from | |
| 167 | + * `data-xs-src` and stripping markers, and #273 is a standing reminder | |
| 168 | + * that regex-editing these attributes in reverse goes wrong quietly. | |
| 169 | + * | |
| 170 | + * The pristine tag still carries whatever priority-10 filters did to | |
| 171 | + * it, so only OUR changes are dropped. (#469) | |
| 172 | + * | |
| 173 | + * @param string $tag | |
| 174 | + * @param string $handle | |
| 175 | + * @param string $src | |
| 176 | + */ | |
| 177 | + public static function revert_late_marked_tag( $tag, $handle, $src ): string { | |
| 178 | + if ( ! is_string( $tag ) || '' === $tag || ! self::tag_opts_out( $tag ) ) { | |
| 179 | + return (string) $tag; | |
| 180 | + } | |
| 181 | + $handle = (string) $handle; | |
| 182 | + $pristine = isset( self::$pristine_tag[ $handle ] ) ? self::$pristine_tag[ $handle ] : ''; | |
| 183 | + if ( '' !== $pristine && $pristine !== $tag ) { | |
| 184 | + // The marker is on the tag we were handed, not on the snapshot, | |
| 185 | + // so carry it — and everything else the late filter set in the | |
| 186 | + // same pass — over. A consumer reading the rendered HTML (or | |
| 187 | + // our own buffer passes) must still see the opt-out it asked | |
| 188 | + // for. | |
| 189 | + $tag = self::copy_late_attributes( $tag, $pristine ); | |
| 190 | + } | |
| 191 | + // The snapshot was taken on `script_loader_tag`, by which point | |
| 192 | + // `script_loader_src` (priority 10) had ALREADY swapped in the | |
| 193 | + // hashed cache URL — so reverting the tag alone still leaves the | |
| 194 | + // minified src behind, which is the half the client actually | |
| 195 | + // reported. Undo that here too, using the URL rewrite_script() | |
| 196 | + // recorded. (#469) | |
| 197 | + return self::restore_marked_script_src( $tag, $handle, self::current_src( $tag, (string) $src ) ); | |
| 198 | + } | |
| 199 | + | |
| 200 | + /** | |
| 201 | + * The src currently on a tag, falling back to the one WordPress passed. | |
| 202 | + * | |
| 203 | + * After a revert the tag carries the snapshot's src, which is not | |
| 204 | + * necessarily the `$src` argument this late in the chain. | |
| 205 | + * | |
| 206 | + * @param string $tag Tag to read. | |
| 207 | + * @param string $fallback Value to use when the tag has no src. | |
| 208 | + */ | |
| 209 | + private static function current_src( string $tag, string $fallback ): string { | |
| 210 | + $open = self::open_tag_offsets( $tag ); | |
| 211 | + if ( null !== $open | |
| 212 | + && preg_match( '#(?<![-\w])src\s*=\s*["\']([^"\']*)["\']#i', $open['attrs'], $m ) ) { | |
| 213 | + return $m[1]; | |
| 214 | + } | |
| 215 | + return $fallback; | |
| 216 | + } | |
| 217 | + | |
| 218 | + /** | |
| 219 | + * Carry the attributes a late filter added onto the snapshot tag. | |
| 220 | + * | |
| 221 | + * Copying only `data-no-*` would silently drop the rest of what the | |
| 222 | + * stamper set in the same pass. Borlabs adds `data-cfasync="false"` | |
| 223 | + * alongside its markers — the attribute that keeps Cloudflare Rocket | |
| 224 | + * Loader off the consent config, i.e. the same class of breakage this | |
| 225 | + * fix exists to prevent, reintroduced by the fix itself. So diff the | |
| 226 | + * attribute names and bring over every one the snapshot lacks. | |
| 227 | + * | |
| 228 | + * Our own transform markers are excluded: they are what we are | |
| 229 | + * reverting, and re-adding `data-xs-delay` would re-delay the script. | |
| 230 | + * | |
| 231 | + * @param string $from Tag as the late filter left it. | |
| 232 | + * @param string $to Snapshot tag to stamp onto. | |
| 233 | + */ | |
| 234 | + private static function copy_late_attributes( string $from, string $to ): string { | |
| 235 | + $late_tags = self::open_tags( $from ); | |
| 236 | + $to_tags = self::open_tags( $to ); | |
| 237 | + if ( empty( $late_tags ) || empty( $to_tags ) || count( $late_tags ) !== count( $to_tags ) ) { | |
| 238 | + // Counts differ when a stamper/blocker injected or replaced a | |
| 239 | + // tag inside the concatenated string, or a transform dropped an | |
| 240 | + // inline block. Positional pairing is meaningless then — but | |
| 241 | + // returning the bare snapshot would silently strip the opt-out, | |
| 242 | + // and the buffer passes would re-optimize an unmarked tag: #469 | |
| 243 | + // again, on the mismatch path. Over-marking merely leaves a tag | |
| 244 | + // unoptimized, so stamp the protective attributes onto every | |
| 245 | + // snapshot tag instead. (#470) | |
| 246 | + return self::stamp_protective_attributes( $from, $to, $to_tags ); | |
| 247 | + } | |
| 248 | + // Pair the tags positionally and stamp each one from its own | |
| 249 | + // counterpart. A stamper runs over the whole concatenated string | |
| 250 | + // and may mark several of the tags in it; collapsing that onto one | |
| 251 | + // tag would strip the opt-out from the others, and the buffer | |
| 252 | + // passes re-test `tag_opts_out()` per tag, so an unmarked sibling | |
| 253 | + // is free to be re-optimized downstream — #469 again, one pass | |
| 254 | + // later. (#469) | |
| 255 | + $out = $to; | |
| 256 | + // Right to left: an earlier splice would shift every later offset. | |
| 257 | + for ( $i = count( $to_tags ) - 1; $i >= 0; $i-- ) { | |
| 258 | + $add = self::late_attribute_delta( $late_tags[ $i ]['attrs'], $to_tags[ $i ]['attrs'] ); | |
| 259 | + if ( '' !== $add ) { | |
| 260 | + $out = substr_replace( $out, $add, $to_tags[ $i ]['attrs_end'], 0 ); | |
| 261 | + } | |
| 262 | + } | |
| 263 | + return $out; | |
| 264 | + } | |
| 265 | + | |
| 266 | + /** | |
| 267 | + * Fallback when the late tag and the snapshot cannot be paired | |
| 268 | + * positionally: copy only the attributes that protect the script from | |
| 269 | + * optimizers — the opt-out markers plus `data-cfasync` — onto every | |
| 270 | + * snapshot tag missing them. Values are taken as the stamper wrote | |
| 271 | + * them on the late tag. (#470) | |
| 272 | + * | |
| 273 | + * @param string $from Tag as the late filter left it. | |
| 274 | + * @param string $to Snapshot tag to stamp onto. | |
| 275 | + * @param array $to_tags open_tags() result for $to. | |
| 276 | + */ | |
| 277 | + private static function stamp_protective_attributes( string $from, string $to, array $to_tags ): string { | |
| 278 | + $protect = array(); | |
| 279 | + foreach ( array( 'data-no-optimize', 'data-no-minify', 'data-cfasync' ) as $name ) { | |
| 280 | + if ( preg_match( | |
| 281 | + '#\s(' . preg_quote( $name, '#' ) . ')(\s*=\s*(?:"[^"]*"|\'[^\']*\'|[^\s>]*))?#i', | |
| 282 | + $from, | |
| 283 | + $m | |
| 284 | + ) ) { | |
| 285 | + $protect[ $name ] = ' ' . $name . ( isset( $m[2] ) ? $m[2] : '' ); | |
| 286 | + } | |
| 287 | + } | |
| 288 | + if ( empty( $protect ) ) { | |
| 289 | + return $to; | |
| 290 | + } | |
| 291 | + $out = $to; | |
| 292 | + // Right to left: an earlier splice would shift every later offset. | |
| 293 | + for ( $i = count( $to_tags ) - 1; $i >= 0; $i-- ) { | |
| 294 | + $add = ''; | |
| 295 | + foreach ( $protect as $name => $attr ) { | |
| 296 | + if ( ! preg_match( '#\s' . preg_quote( $name, '#' ) . '\b#i', $to_tags[ $i ]['attrs'] ) ) { | |
| 297 | + $add .= $attr; | |
| 298 | + } | |
| 299 | + } | |
| 300 | + if ( '' !== $add ) { | |
| 301 | + $out = substr_replace( $out, $add, $to_tags[ $i ]['attrs_end'], 0 ); | |
| 302 | + } | |
| 303 | + } | |
| 304 | + return $out; | |
| 305 | + } | |
| 306 | + | |
| 307 | + /** | |
| 308 | + * The attributes present on the late tag but not the snapshot, minus | |
| 309 | + * the ones our own transforms add. | |
| 310 | + * | |
| 311 | + * @param string $late_attrs Attribute string from the transformed tag. | |
| 312 | + * @param string $to_attrs Attribute string from the snapshot tag. | |
| 313 | + */ | |
| 314 | + private static function late_attribute_delta( string $late_attrs, string $to_attrs ): string { | |
| 315 | + $pattern = '#\s([-\w:]+)(?:\s*=\s*(?:"[^"]*"|\'[^\']*\'|[^\s>]*))?#'; | |
| 316 | + if ( ! preg_match_all( $pattern, $late_attrs, $late, PREG_SET_ORDER ) ) { | |
| 317 | + return ''; | |
| 318 | + } | |
| 319 | + $have = array(); | |
| 320 | + if ( preg_match_all( $pattern, $to_attrs, $existing, PREG_SET_ORDER ) ) { | |
| 321 | + foreach ( $existing as $attr ) { | |
| 322 | + $have[ strtolower( $attr[1] ) ] = true; | |
| 323 | + } | |
| 324 | + } | |
| 325 | + $add = ''; | |
| 326 | + foreach ( $late as $attr ) { | |
| 327 | + $name = strtolower( $attr[1] ); | |
| 328 | + if ( isset( $have[ $name ] ) || in_array( $name, self::OUR_TRANSFORM_ATTRS, true ) ) { | |
| 329 | + continue; | |
| 330 | + } | |
| 331 | + if ( 0 === strpos( $name, 'data-xs-' ) ) { | |
| 332 | + continue; | |
| 333 | + } | |
| 334 | + $add .= $attr[0]; | |
| 335 | + } | |
| 336 | + return $add; | |
| 337 | + } | |
| 338 | + | |
| 339 | + /** | |
| 340 | + * Attributes our own transforms add. Copying any of these from the | |
| 341 | + * transformed tag back onto the snapshot would re-apply the very | |
| 342 | + * transform we are undoing: | |
| 343 | + * | |
| 344 | + * defer/async — defer_script_tag() | |
| 345 | + * type — delay_script_tag() parks an inline block as | |
| 346 | + * text/xspeed-delayed; a type the author set is on the | |
| 347 | + * snapshot already and matches by name before we get here | |
| 348 | + * src — belongs to the snapshot, never to the late tag | |
| 349 | + * | |
| 350 | + * `data-xs-*` is handled by prefix separately. (#469) | |
| 351 | + */ | |
| 352 | + private const OUR_TRANSFORM_ATTRS = array( 'defer', 'async', 'type', 'src' ); | |
| 353 | + | |
| 354 | + /** | |
| 355 | + * Locate the opening `<script>` that carries the src, falling back to | |
| 356 | + * the last one when none does. | |
| 357 | + * | |
| 358 | + * WP_Scripts::do_item() hands `script_loader_tag` the concatenation of | |
| 359 | + * before_inline + external + after_inline, so the FIRST `<script` is | |
| 360 | + * routinely an inline block rather than the asset — the same trap | |
| 361 | + * #234 fixed for defer and #273 for delay. Scanning attribute-wise | |
| 362 | + * also means a quoted value containing `>` (an `onerror` guard, a JSON | |
| 363 | + * payload) cannot truncate the tag the way `[^>]*` did. (#469) | |
| 364 | + * | |
| 365 | + * @param string $tag Full tag string. | |
| 366 | + * @return array{attrs:string,attrs_end:int}|null | |
| 367 | + */ | |
| 368 | + private static function open_tag_offsets( string $tag ): ?array { | |
| 369 | + $tags = self::open_tags( $tag ); | |
| 370 | + $fallback = null; | |
| 371 | + foreach ( $tags as $found ) { | |
| 372 | + if ( preg_match( '#(?<![-\w])src\s*=#i', $found['attrs'] ) ) { | |
| 373 | + return $found; | |
| 374 | + } | |
| 375 | + $fallback = $found; | |
| 376 | + } | |
| 377 | + return $fallback; | |
| 378 | + } | |
| 379 | + | |
| 380 | + /** | |
| 381 | + * Every well-formed opening `<script>` in the string, in order. | |
| 382 | + * | |
| 383 | + * @param string $tag Full tag string. | |
| 384 | + * @return array<int,array{attrs:string,attrs_end:int}> | |
| 385 | + */ | |
| 386 | + private static function open_tags( string $tag ): array { | |
| 387 | + if ( ! preg_match_all( '#<script\b#i', $tag, $m, PREG_OFFSET_CAPTURE ) ) { | |
| 388 | + return array(); | |
| 389 | + } | |
| 390 | + $found = array(); | |
| 391 | + foreach ( $m[0] as $hit ) { | |
| 392 | + $start = (int) $hit[1] + strlen( $hit[0] ); | |
| 393 | + $end = self::scan_open_tag_end( $tag, $start ); | |
| 394 | + if ( null === $end ) { | |
| 395 | + continue; | |
| 396 | + } | |
| 397 | + $found[] = array( | |
| 398 | + 'attrs' => substr( $tag, $start, $end - $start ), | |
| 399 | + 'attrs_end' => $end, | |
| 400 | + ); | |
| 401 | + } | |
| 402 | + return $found; | |
| 403 | + } | |
| 404 | + | |
| 405 | + /** | |
| 406 | + * Offset of the `>` closing an opening tag, skipping any that sit | |
| 407 | + * inside a quoted attribute value. Null when the tag is unterminated. | |
| 408 | + * | |
| 409 | + * @param string $tag Full tag string. | |
| 410 | + * @param int $offset Index just past `<script`. | |
| 411 | + */ | |
| 412 | + private static function scan_open_tag_end( string $tag, int $offset ): ?int { | |
| 413 | + $len = strlen( $tag ); | |
| 414 | + $quote = ''; | |
| 415 | + for ( $i = $offset; $i < $len; $i++ ) { | |
| 416 | + $char = $tag[ $i ]; | |
| 417 | + if ( '' !== $quote ) { | |
| 418 | + if ( $char === $quote ) { | |
| 419 | + $quote = ''; | |
| 420 | + } | |
| 421 | + continue; | |
| 422 | + } | |
| 423 | + if ( '"' === $char || "'" === $char ) { | |
| 424 | + $quote = $char; | |
| 425 | + continue; | |
| 426 | + } | |
| 427 | + if ( '>' === $char ) { | |
| 428 | + // A self-closing `/>` keeps the slash out of the attributes. | |
| 429 | + return ( $i > $offset && '/' === $tag[ $i - 1 ] ) ? $i - 1 : $i; | |
| 430 | + } | |
| 431 | + } | |
| 432 | + return null; | |
| 433 | + } | |
| 434 | + | |
| 435 | + /** | |
| 436 | + * Filter: `script_loader_tag`, priority 15 — undo the minify-cache | |
| 437 | + * rewrite for a script whose printed tag opts out. | |
| 438 | + * | |
| 439 | + * The src rewrite happens on `script_loader_src` (priority 10), long | |
| 440 | + * before any plugin's own `script_loader_tag` filter can stamp | |
| 441 | + * `data-no-minify` onto the tag — so the marker arrived too late to | |
| 442 | + * prevent the rewrite. This runs after the filters that stamp at the | |
| 443 | + * default priority 10 and swaps the hashed cache URL back to the | |
| 444 | + * recorded original. A marker stamped later than our transforms is | |
| 445 | + * caught by revert_late_marked_tag() in the late pass instead. (#469) | |
| 446 | + * | |
| 447 | + * @param string $tag | |
| 448 | + * @param string $handle | |
| 449 | + * @param string $src | |
| 450 | + */ | |
| 451 | + public static function restore_marked_script_src( $tag, $handle, $src ): string { | |
| 452 | + if ( ! is_string( $tag ) || '' === $tag || ! self::tag_opts_out( $tag ) ) { | |
| 453 | + return (string) $tag; | |
| 454 | + } | |
| 455 | + $original = self::original_src( (string) $handle ); | |
| 456 | + if ( '' === $original || '' === (string) $src || false === strpos( $tag, (string) $src ) ) { | |
| 457 | + return $tag; | |
| 458 | + } | |
| 459 | + return str_replace( (string) $src, $original, $tag ); | |
| 460 | + } | |
| 461 | + | |
| 462 | + /** | |
| 86 | 463 | * Does a user-supplied target match this script? |
| 87 | 464 | * |
| 88 | 465 | * A target is either a script handle (exact) or a URL substring. The |
| 89 | 466 | * URL is checked against BOTH the current src and the pre-minify src, |
| @@ -138,8 +515,22 @@ | ||
| 138 | 515 | } |
| 139 | 516 | if ( self::is_excluded_script( (string) $handle, (string) $src ) ) { |
| 140 | 517 | return $tag; |
| 141 | 518 | } |
| 519 | + // The tag itself asked to be left alone. (#456) | |
| 520 | + if ( self::tag_opts_out( $tag ) ) { | |
| 521 | + return $tag; | |
| 522 | + } | |
| 523 | + // Inline code elsewhere on the page reads this handle (or something | |
| 524 | + // it depends on). Inline blocks never defer, so deferring this one | |
| 525 | + // would run the consumer first. Defer only — delay is an opt-in | |
| 526 | + // target list, where the user has named the script deliberately. | |
| 527 | + if ( isset( self::inline_bound_handles()[ (string) $handle ] ) ) { | |
| 528 | + return $tag; | |
| 529 | + } | |
| 530 | + // NB: is_protected_from_bundling() is the same two rules in one call | |
| 531 | + // for the combiner; the split here is deliberate, since the | |
| 532 | + // exclusion check above already ran and short-circuits earlier. | |
| 142 | 533 | if ( false !== stripos( $tag, ' defer' ) || false !== stripos( $tag, ' async' ) ) { |
| 143 | 534 | return $tag; |
| 144 | 535 | } |
| 145 | 536 | // Target the <script> that actually carries a src, NOT simply the |
| @@ -177,17 +568,32 @@ | ||
| 177 | 568 | } |
| 178 | 569 | if ( self::is_excluded_script( (string) $handle, (string) $src ) ) { |
| 179 | 570 | return $tag; |
| 180 | 571 | } |
| 572 | + // The tag itself asked to be left alone. (#456) | |
| 573 | + if ( self::tag_opts_out( $tag ) ) { | |
| 574 | + return $tag; | |
| 575 | + } | |
| 181 | 576 | if ( ! self::is_delay_target( (string) $handle, (string) $src ) ) { |
| 182 | 577 | return $tag; |
| 183 | 578 | } |
| 579 | + // Inline code elsewhere on the page reads this handle (or something | |
| 580 | + // it depends on) — same registry walk defer uses. Delaying it runs | |
| 581 | + // the consumer at parse time against a global that arrives on first | |
| 582 | + // interaction: `wp_add_inline_script( 'jquery-ui-core', | |
| 583 | + // 'jQuery.uiBackCompat…', 'before' )` throws "jQuery is not defined" | |
| 584 | + // the moment jquery-core is delayed. A handle the user NAMED in | |
| 585 | + // delay_js_targets is still delayed — an explicit entry is the user | |
| 586 | + // saying they know the inline consumer is safe to break or absent. | |
| 587 | + if ( isset( self::inline_bound_handles()[ (string) $handle ] ) | |
| 588 | + && ! self::is_user_named_target( (string) $handle, (string) $src ) ) { | |
| 589 | + return $tag; | |
| 590 | + } | |
| 184 | 591 | // A non-executable type means this tag is data, or is being held by |
| 185 | 592 | // somebody else on purpose. The buffer pass has always checked this; |
| 186 | 593 | // the enqueue path did not, so a consent-blocked or JSON-carrying |
| 187 | 594 | // handle could still be rewritten here. (#274) |
| 188 | - if ( preg_match( '#\btype\s*=\s*(["\'])(.*?)\1#is', $tag, $type_m ) | |
| 189 | - && in_array( strtolower( trim( $type_m[2] ) ), self::NON_EXECUTABLE_TYPES, true ) ) { | |
| 595 | + if ( in_array( self::extract_type( $tag ), self::NON_EXECUTABLE_TYPES, true ) ) { | |
| 190 | 596 | return $tag; |
| 191 | 597 | } |
| 192 | 598 | // src= variant: swap src → data-xs-src and add data-xs-delay marker. |
| 193 | 599 | if ( '' !== (string) $src ) { |
| @@ -215,16 +621,25 @@ | ||
| 215 | 621 | $tag, |
| 216 | 622 | 1 |
| 217 | 623 | ); |
| 218 | 624 | } |
| 219 | - // Inline script: change type to text/plain so the browser | |
| 220 | - // doesn't execute, mark for bootstrap rewriter. | |
| 221 | - return (string) preg_replace( | |
| 625 | + // Inline script: change type to text/xspeed-delayed so the browser | |
| 626 | + // doesn't execute, mark for bootstrap rewriter. Any existing type | |
| 627 | + // is REPLACED, not appended-after: HTML keeps an attribute's first | |
| 628 | + // occurrence, so a snippet carrying its own `type="text/javascript"` | |
| 629 | + // would win over a marker appended behind it and keep executing. | |
| 630 | + // A non-default original type is stashed in data-xs-type so the | |
| 631 | + // bootstrap can restore it on replay (#274 — type is what a script | |
| 632 | + // IS; a parked `type="module"` must come back as a module). | |
| 633 | + $tag = (string) preg_replace_callback( | |
| 222 | 634 | '#<script\b([^>]*)>#i', |
| 223 | - '<script$1 type="text/xspeed-delayed" data-xs-delay="1">', | |
| 635 | + static function ( array $m ): string { | |
| 636 | + return '<script' . self::park_type_attrs( $m[1] ) . '>'; | |
| 637 | + }, | |
| 224 | 638 | $tag, |
| 225 | 639 | 1 |
| 226 | 640 | ); |
| 641 | + return $tag; | |
| 227 | 642 | } |
| 228 | 643 | |
| 229 | 644 | /** |
| 230 | 645 | * Script types the buffer pass must never touch. `<script>` carries |
| @@ -248,8 +663,74 @@ | ||
| 248 | 663 | 'text/plain', |
| 249 | 664 | ); |
| 250 | 665 | |
| 251 | 666 | /** |
| 667 | + * The `type` attribute, quoted OR unquoted, anchored to attribute | |
| 668 | + * position — a required leading whitespace, never a bare `\b`. | |
| 669 | + * | |
| 670 | + * The anchoring matters twice over. `\btype` also matches the tail of | |
| 671 | + * any hyphenated `data-…-type` attribute (a `-` is a non-word char, so | |
| 672 | + * the boundary sits inside the name — the same #273 class as `src`), | |
| 673 | + * and it matches a `type=` sitting INSIDE another attribute's value | |
| 674 | + * (`onload="this.type='done'"`). Requiring whitespace before the name | |
| 675 | + * rules both out: attributes are whitespace-separated, while `.type` | |
| 676 | + * and `-type` never are. The unquoted branch exists because | |
| 677 | + * `type=text/javascript` is valid HTML: a quoted-only pattern left it | |
| 678 | + * standing, the parking type appended after it lost the | |
| 679 | + * first-occurrence race, and the snippet executed immediately AND | |
| 680 | + * replayed on interaction — every vendor event fired twice. | |
| 681 | + */ | |
| 682 | + private const TYPE_ATTR_RE = '#\stype\s*=\s*(?:(["\'])(.*?)\1|([^\s>]+))#is'; | |
| 683 | + | |
| 684 | + /** | |
| 685 | + * `type` values a parked tag need not remember: the replay default is | |
| 686 | + * already JavaScript, so stashing these would only fatten the markup. | |
| 687 | + */ | |
| 688 | + private const DEFAULT_JS_TYPES = array( | |
| 689 | + 'text/javascript', | |
| 690 | + 'application/javascript', | |
| 691 | + ); | |
| 692 | + | |
| 693 | + /** | |
| 694 | + * Read a tag's `type` attribute value, lowercased and trimmed. | |
| 695 | + * | |
| 696 | + * @param string $haystack Full tag or its attribute string. | |
| 697 | + * @return string '' when no type attribute is present. | |
| 698 | + */ | |
| 699 | + private static function extract_type( string $haystack ): string { | |
| 700 | + if ( ! preg_match( self::TYPE_ATTR_RE, $haystack, $m ) ) { | |
| 701 | + return ''; | |
| 702 | + } | |
| 703 | + $value = ( isset( $m[3] ) && '' !== $m[3] ) ? $m[3] : $m[2]; | |
| 704 | + return strtolower( trim( $value ) ); | |
| 705 | + } | |
| 706 | + | |
| 707 | + /** | |
| 708 | + * Rewrite an inline tag's attribute string for parking: strip its own | |
| 709 | + * `type`, stash a non-default one in `data-xs-type` (the bootstrap | |
| 710 | + * restores it on replay, so a parked `type="module"` comes back as a | |
| 711 | + * module rather than a classic script — #274), and append the parking | |
| 712 | + * marker pair. | |
| 713 | + * | |
| 714 | + * @param string $attrs Raw attribute string (everything between | |
| 715 | + * `<script` and `>`). | |
| 716 | + */ | |
| 717 | + private static function park_type_attrs( string $attrs ): string { | |
| 718 | + $orig = self::extract_type( $attrs ); | |
| 719 | + $attrs = (string) preg_replace( self::TYPE_ATTR_RE, '', $attrs ); | |
| 720 | + $stash = ''; | |
| 721 | + if ( '' !== $orig && ! in_array( $orig, self::DEFAULT_JS_TYPES, true ) ) { | |
| 722 | + // MIME-ish charset only — a type value is never markup, and this | |
| 723 | + // string is re-emitted inside a double-quoted attribute. | |
| 724 | + $orig = (string) preg_replace( '#[^a-z0-9/+.\-]#', '', $orig ); | |
| 725 | + if ( '' !== $orig ) { | |
| 726 | + $stash = ' data-xs-type="' . $orig . '"'; | |
| 727 | + } | |
| 728 | + } | |
| 729 | + return $attrs . $stash . ' type="text/xspeed-delayed" data-xs-delay="1"'; | |
| 730 | + } | |
| 731 | + | |
| 732 | + /** | |
| 252 | 733 | * URL fragments that must keep a live src no matter what. The enqueue |
| 253 | 734 | * path guards these by handle (ALWAYS_EXCLUDED_HANDLES), but a buffer |
| 254 | 735 | * pass only ever sees a URL, so the same protection is re-expressed |
| 255 | 736 | * here. Without this the admin bundle could be delayed on a frontend |
| @@ -301,8 +782,13 @@ | ||
| 301 | 782 | if ( false !== stripos( $tag, 'data-xs-delay' ) || false !== stripos( $tag, 'data-xs-src' ) ) { |
| 302 | 783 | return $tag; |
| 303 | 784 | } |
| 304 | 785 | |
| 786 | + // The tag itself asked to be left alone. (#456) | |
| 787 | + if ( self::tag_opts_out( $tag ) ) { | |
| 788 | + return $tag; | |
| 789 | + } | |
| 790 | + | |
| 305 | 791 | // No src → inline code. The enqueue path owns those; a |
| 306 | 792 | // buffer rewrite here would have to reason about execution |
| 307 | 793 | // order it cannot see. |
| 308 | 794 | // `(?<![-\w])` not `\b` — see the note on the enqueue-path |
| @@ -315,13 +801,10 @@ | ||
| 315 | 801 | } |
| 316 | 802 | $src = $src_m[2]; |
| 317 | 803 | |
| 318 | 804 | // Data, not code. |
| 319 | - if ( preg_match( '#\btype\s*=\s*(["\'])(.*?)\1#is', $tag, $type_m ) ) { | |
| 320 | - $type = strtolower( trim( $type_m[2] ) ); | |
| 321 | - if ( in_array( $type, self::NON_EXECUTABLE_TYPES, true ) ) { | |
| 322 | - return $tag; | |
| 323 | - } | |
| 805 | + if ( in_array( self::extract_type( $tag ), self::NON_EXECUTABLE_TYPES, true ) ) { | |
| 806 | + return $tag; | |
| 324 | 807 | } |
| 325 | 808 | |
| 326 | 809 | foreach ( self::ALWAYS_EXCLUDED_SRC as $needle ) { |
| 327 | 810 | if ( false !== stripos( $src, $needle ) ) { |
| @@ -328,15 +811,50 @@ | ||
| 328 | 811 | return $tag; |
| 329 | 812 | } |
| 330 | 813 | } |
| 331 | 814 | |
| 332 | - // Buffer-pass tags have no handle — match on URL only. | |
| 333 | - if ( self::is_excluded_script( '', $src ) ) { | |
| 815 | + // Recover the handle from the tag's id before deciding. | |
| 816 | + // | |
| 817 | + // This pass used to pass '' as the handle, on the reasoning | |
| 818 | + // that a tag reaching the buffer was never enqueued and so has | |
| 819 | + // none. That holds for the third-party snippets this pass | |
| 820 | + // exists for — but NOT for enqueued scripts, which also travel | |
| 821 | + // through here, and which WordPress prints with | |
| 822 | + // `id="<handle>-js"`. Passing '' meant every handle-based | |
| 823 | + // exclusion was silently inert at this layer: the user writes | |
| 824 | + // `jquery-core`, the enqueue path honours it, and then the | |
| 825 | + // buffer pass — which only ever compared URLs — delayed the | |
| 826 | + // very script the list was protecting. | |
| 827 | + // | |
| 828 | + // That is how a site with jquery-core AND jquery-migrate | |
| 829 | + // excluded still shipped jQuery delayed while migrate loaded | |
| 830 | + // normally, and every inline `jQuery(...)` on the page threw | |
| 831 | + // "jQuery is not defined". The two behaved differently for no | |
| 832 | + // reason a user could see, which is what made it look like a | |
| 833 | + // matching quirk rather than a whole layer ignoring the list. | |
| 834 | + $tag_handle = ''; | |
| 835 | + if ( preg_match( '#\sid\s*=\s*(["\'])(.*?)\1#i', $tag, $id_m ) ) { | |
| 836 | + // WP appends `-js`; anything else is somebody's own id and | |
| 837 | + // is still worth matching literally. | |
| 838 | + $tag_handle = (string) preg_replace( '/-js$/', '', $id_m[2] ); | |
| 839 | + } | |
| 840 | + | |
| 841 | + if ( self::is_excluded_script( $tag_handle, $src ) ) { | |
| 334 | 842 | return $tag; |
| 335 | 843 | } |
| 336 | - if ( ! self::is_delay_target( '', $src ) ) { | |
| 844 | + if ( ! self::is_delay_target( $tag_handle, $src ) ) { | |
| 337 | 845 | return $tag; |
| 338 | 846 | } |
| 847 | + // Mirror of the enqueue-path guard: a handle that inline code | |
| 848 | + // reads stays eager unless the user named it. wp_scripts() | |
| 849 | + // is still populated at xspeed_cache_final_html time on a | |
| 850 | + // MISS, so the registry walk is consultable here too; an | |
| 851 | + // unrecoverable handle ('') simply never matches the set. | |
| 852 | + if ( '' !== $tag_handle | |
| 853 | + && isset( self::inline_bound_handles()[ $tag_handle ] ) | |
| 854 | + && ! self::is_user_named_target( $tag_handle, $src ) ) { | |
| 855 | + return $tag; | |
| 856 | + } | |
| 339 | 857 | |
| 340 | 858 | return (string) preg_replace( |
| 341 | 859 | '#(?<![-\w])src\s*=\s*(["\'][^"\']*["\'])#i', |
| 342 | 860 | 'data-xs-src=$1 data-xs-delay="1"', |
| @@ -348,8 +866,119 @@ | ||
| 348 | 866 | ); |
| 349 | 867 | } |
| 350 | 868 | |
| 351 | 869 | /** |
| 870 | + * Delay inline vendor snippets that reference a known third-party host. | |
| 871 | + * | |
| 872 | + * The pass above rewrites `src` and deliberately leaves inline code | |
| 873 | + * alone — but the OFFICIAL install for Clarity, GA, GTM and the Meta | |
| 874 | + * pixel is an inline loader (`(function(c,l,a,r,i,t,y){…t.src=…})`) | |
| 875 | + * with no `src` attribute at all. That snippet executes on every page | |
| 876 | + * load, fetches the vendor bundle inside the measurement window, and | |
| 877 | + * puts the one host whose Cache-Control the site cannot set straight | |
| 878 | + * into the cache-policy and TBT audits. Delaying the enqueue path and | |
| 879 | + * the raw-src path while this runs untouched is delaying everything | |
| 880 | + * except the tag the feature exists for. | |
| 881 | + * | |
| 882 | + * The judgment call is the same one KNOWN_THIRD_PARTY_SRC already | |
| 883 | + * makes: an inline body that names one of those hosts is that vendor's | |
| 884 | + * loader or its config — never something first-party code holds a | |
| 885 | + * synchronous reference to. The body is the haystack for the user's | |
| 886 | + * exclusion and target lists too, so the same fragment that protects a | |
| 887 | + * `src` tag protects its inline install. | |
| 888 | + * | |
| 889 | + * `document.write` bodies are skipped outright: replayed after the | |
| 890 | + * parser has closed the document, a delayed write would replace the | |
| 891 | + * page rather than add to it. | |
| 892 | + * | |
| 893 | + * @param string $html Complete page HTML. | |
| 894 | + */ | |
| 895 | + public static function delay_inline_snippets( $html ): string { | |
| 896 | + if ( ! is_string( $html ) || '' === $html ) { | |
| 897 | + return (string) $html; | |
| 898 | + } | |
| 899 | + if ( self::skip_in_non_frontend_context() ) { | |
| 900 | + return $html; | |
| 901 | + } | |
| 902 | + $opts = self::opts(); | |
| 903 | + if ( empty( $opts['delay_js'] ) ) { | |
| 904 | + return $html; | |
| 905 | + } | |
| 906 | + | |
| 907 | + $out = preg_replace_callback( | |
| 908 | + '#<script\b([^>]*)>(.*?)</script>#is', | |
| 909 | + static function ( array $m ): string { | |
| 910 | + list( $whole, $attrs, $body ) = $m; | |
| 911 | + | |
| 912 | + if ( '' === trim( $body ) ) { | |
| 913 | + return $whole; | |
| 914 | + } | |
| 915 | + | |
| 916 | + // The tag itself asked to be left alone. (#456) | |
| 917 | + if ( self::tag_opts_out( $attrs ) ) { | |
| 918 | + return $whole; | |
| 919 | + } | |
| 920 | + | |
| 921 | + // Our own replay bootstrap. Its body quotes the delay | |
| 922 | + // machinery's own strings, so a pathological user target | |
| 923 | + // fragment could match it — and a parked bootstrap means | |
| 924 | + // nothing on the page ever replays. | |
| 925 | + if ( false !== stripos( $attrs, 'xspeed-delay-bootstrap' ) ) { | |
| 926 | + return $whole; | |
| 927 | + } | |
| 928 | + | |
| 929 | + // Already marked, or a real src= — the src passes own those. | |
| 930 | + // `(?<![-\w])` for the same reason as above: `data-cmplz-src` | |
| 931 | + // must not read as a src. (#273) | |
| 932 | + if ( false !== stripos( $attrs, 'data-xs-delay' ) || false !== stripos( $attrs, 'data-xs-src' ) ) { | |
| 933 | + return $whole; | |
| 934 | + } | |
| 935 | + if ( preg_match( '#(?<![-\w])src\s*=\s*(["\']).*?\1#is', $attrs ) ) { | |
| 936 | + return $whole; | |
| 937 | + } | |
| 938 | + | |
| 939 | + // Data, a module map, or a consent manager's parked tag. | |
| 940 | + if ( in_array( self::extract_type( $attrs ), self::NON_EXECUTABLE_TYPES, true ) ) { | |
| 941 | + return $whole; | |
| 942 | + } | |
| 943 | + | |
| 944 | + // A delayed document.write replays after the document has | |
| 945 | + // closed and replaces the page. Never delay one. | |
| 946 | + if ( false !== stripos( $body, 'document.write' ) ) { | |
| 947 | + return $whole; | |
| 948 | + } | |
| 949 | + | |
| 950 | + // The body stands in for the URL in the lists the src passes | |
| 951 | + // consult — but NOT via is_delay_target(), whose empty-list | |
| 952 | + // default is "delay everything". That default is right for a | |
| 953 | + // tag with a URL and catastrophic here: it would park every | |
| 954 | + // inline script on the page. Inline code is delayed only on a | |
| 955 | + // positive identification — the body names a known vendor | |
| 956 | + // host, or a fragment the user targeted — and the exclusion | |
| 957 | + // list still wins first. | |
| 958 | + if ( self::is_excluded_script( '', $body ) ) { | |
| 959 | + return $whole; | |
| 960 | + } | |
| 961 | + if ( ! self::matches_known_third_party( $body ) && ! self::matches_user_targets( $body ) ) { | |
| 962 | + return $whole; | |
| 963 | + } | |
| 964 | + | |
| 965 | + // Replace — not append — any existing type. Attributes keep | |
| 966 | + // their FIRST occurrence in HTML, so appending the parking | |
| 967 | + // type after the snippet's own `type="text/javascript"` | |
| 968 | + // would leave the original executable. A non-default type is | |
| 969 | + // stashed in data-xs-type for the bootstrap to restore. | |
| 970 | + return '<script' . self::park_type_attrs( $attrs ) . '>' . $body . '</script>'; | |
| 971 | + }, | |
| 972 | + $html | |
| 973 | + ); | |
| 974 | + // A PCRE failure (backtrack limit on a huge inline body) returns | |
| 975 | + // null — and casting that to '' would serve AND cache a blank page. | |
| 976 | + // The unrewritten original is always the safe fallback. | |
| 977 | + return null === $out ? $html : $out; | |
| 978 | + } | |
| 979 | + | |
| 980 | + /** | |
| 352 | 981 | * Inline bootstrap that flips delayed scripts on the first user |
| 353 | 982 | * interaction. Printed once on wp_footer priority 1000. |
| 354 | 983 | */ |
| 355 | 984 | public static function print_delay_bootstrap(): void { |
| @@ -382,11 +1011,30 @@ | ||
| 382 | 1011 | events.forEach(function(e){window.removeEventListener(e,load,{passive:true,capture:true});}); |
| 383 | 1012 | var delayed=document.querySelectorAll('script[data-xs-delay]'); |
| 384 | 1013 | delayed.forEach(function(s){ |
| 385 | 1014 | var n=document.createElement('script'); |
| 1015 | + // A dynamically-created script is async by default, so replayed | |
| 1016 | + // EXTERNALS would race each other; async=false restores document | |
| 1017 | + // order among the externals. Narrower guarantee, stated plainly: | |
| 1018 | + // a replayed INLINE script still executes synchronously at its | |
| 1019 | + // replaceChild, i.e. possibly before an earlier external has | |
| 1020 | + // finished LOADING — so an inline consumer of a delayed external | |
| 1021 | + // is only safe when both were delayed by explicit user targeting | |
| 1022 | + // (the inline-bound guard keeps the implicit case eager). | |
| 1023 | + n.async=false; | |
| 1024 | + // Nonce hiding: a connected element's nonce CONTENT attribute reads | |
| 1025 | + // as "", so copying it via the attribute loop would hand the clone | |
| 1026 | + // an empty nonce and a nonce-based CSP would block the replay. The | |
| 1027 | + // IDL property still carries the real value. | |
| 1028 | + if(s.nonce){n.nonce=s.nonce;} | |
| 386 | 1029 | Array.prototype.slice.call(s.attributes).forEach(function(a){ |
| 387 | 1030 | if(a.name==='data-xs-src'){n.setAttribute('src',a.value);return;} |
| 388 | 1031 | if(a.name==='data-xs-delay')return; |
| 1032 | + if(a.name==='nonce')return; | |
| 1033 | + // A parked inline tag's ORIGINAL type (module, mostly) rides in | |
| 1034 | + // data-xs-type — restore it, or the replay runs a module as a | |
| 1035 | + // classic script and its imports throw. (#274) | |
| 1036 | + if(a.name==='data-xs-type'){n.setAttribute('type',a.value);return;} | |
| 389 | 1037 | // `type` is what a script IS, not decoration, so it is carried over |
| 390 | 1038 | // — with ONE exception: our own inline parking marker, which exists |
| 391 | 1039 | // only to stop the browser executing the original and must not be |
| 392 | 1040 | // copied onto the replacement. Dropping type wholesale broke two |
| @@ -501,8 +1149,28 @@ | ||
| 501 | 1149 | } |
| 502 | 1150 | if ( preg_match( '#\bmedia\s*=\s*(["\'])\s*print\s*\1#i', $tag ) ) { |
| 503 | 1151 | return $tag; |
| 504 | 1152 | } |
| 1153 | + return self::async_link_markup( $tag ); | |
| 1154 | + } | |
| 1155 | + | |
| 1156 | + /** | |
| 1157 | + * The one place the async-CSS output shape lives: swap the link's media | |
| 1158 | + * to `print`, restore the original media onload, record it in | |
| 1159 | + * `data-xs-async`, and re-emit the untouched tag inside `<noscript>` for | |
| 1160 | + * clients that never run the onload handler. | |
| 1161 | + * | |
| 1162 | + * Shared by the enqueue-path filter above and the raw-tag buffer pass | |
| 1163 | + * below so the two can never drift — Pro's Critical CSS recognises this | |
| 1164 | + * exact marker to avoid double-wrapping, and a second copy of the | |
| 1165 | + * pattern is how that kind of contract quietly breaks. | |
| 1166 | + * | |
| 1167 | + * Callers own every skip decision (markers, onload, non-screen media); | |
| 1168 | + * this helper only produces the markup. | |
| 1169 | + * | |
| 1170 | + * @param string $tag A `<link rel="stylesheet">` tag deemed safe to defer. | |
| 1171 | + */ | |
| 1172 | + private static function async_link_markup( string $tag ): string { | |
| 505 | 1173 | $async = (string) preg_replace_callback( |
| 506 | 1174 | '#\bmedia\s*=\s*(["\'])([^"\']*)\1#i', |
| 507 | 1175 | static function ( $m ) { |
| 508 | 1176 | $orig = $m[2]; |
| @@ -524,8 +1192,191 @@ | ||
| 524 | 1192 | return $async . '<noscript>' . $tag . '</noscript>'; |
| 525 | 1193 | } |
| 526 | 1194 | |
| 527 | 1195 | /** |
| 1196 | + * Stylesheet hosts that serve FONT CSS — small, render-blocking sheets of | |
| 1197 | + * `@font-face` rules. The buffer pass below defers only these: a raw | |
| 1198 | + * cross-origin `<link>` could carry anything, and blindly deferring an | |
| 1199 | + * unknown vendor's layout CSS from the buffer would reintroduce the | |
| 1200 | + * unstyled-flash failure async_style_tag()'s guards exist to prevent. | |
| 1201 | + * Font CSS is the safe subset — text renders in a fallback face and swaps, | |
| 1202 | + * which is exactly what `font-display: swap` does on purpose. | |
| 1203 | + */ | |
| 1204 | + private const FONT_CSS_HOSTS = array( | |
| 1205 | + 'fonts.googleapis.com', | |
| 1206 | + 'fonts.bunny.net', | |
| 1207 | + 'use.typekit.net', | |
| 1208 | + 'p.typekit.net', | |
| 1209 | + 'fonts.cdnfonts.com', | |
| 1210 | + ); | |
| 1211 | + | |
| 1212 | + /** | |
| 1213 | + * The font-CSS host allowlist, filtered and normalised. | |
| 1214 | + * | |
| 1215 | + * @return string[] Lowercase hostnames. | |
| 1216 | + */ | |
| 1217 | + private static function font_css_hosts(): array { | |
| 1218 | + /** | |
| 1219 | + * Hosts whose stylesheet links the async-CSS buffer pass rewrites to | |
| 1220 | + * the non-blocking print → onload pattern. Only font-CSS providers | |
| 1221 | + * belong here: every listed host's sheets are safe to load late | |
| 1222 | + * because they only add `@font-face` rules. | |
| 1223 | + * | |
| 1224 | + * @param string[] $hosts Hostnames (exact match, case-insensitive). | |
| 1225 | + */ | |
| 1226 | + $hosts = (array) apply_filters( 'xspeed_async_css_font_hosts', self::FONT_CSS_HOSTS ); | |
| 1227 | + | |
| 1228 | + return array_map( 'strtolower', array_map( 'strval', $hosts ) ); | |
| 1229 | + } | |
| 1230 | + | |
| 1231 | + /** | |
| 1232 | + * Media values that never apply to a screen paint. A sheet restricted to | |
| 1233 | + * one of these is not render-blocking for screen, so deferring it saves | |
| 1234 | + * nothing — and `print` in particular is either a genuine print sheet or | |
| 1235 | + * somebody's finished async pattern, both of which must be left alone. | |
| 1236 | + */ | |
| 1237 | + private const NON_SCREEN_MEDIA = array( | |
| 1238 | + 'print', | |
| 1239 | + 'speech', | |
| 1240 | + 'aural', | |
| 1241 | + 'braille', | |
| 1242 | + 'embossed', | |
| 1243 | + 'handheld', | |
| 1244 | + 'projection', | |
| 1245 | + 'tty', | |
| 1246 | + 'tv', | |
| 1247 | + ); | |
| 1248 | + | |
| 1249 | + /** | |
| 1250 | + * Filter: `xspeed_cache_final_html` — defer RAW font-CSS stylesheet links | |
| 1251 | + * that never passed through wp_enqueue_style. | |
| 1252 | + * | |
| 1253 | + * `async_style_tag()` hooks `style_loader_tag`, so it only ever sees | |
| 1254 | + * enqueued stylesheets. Themes and font plugins print Google Fonts (and | |
| 1255 | + * Bunny, Typekit, CDNFonts) as literal | |
| 1256 | + * `<link rel="stylesheet" href="https://fonts.googleapis.com/css?family=…">` | |
| 1257 | + * markup in the head — on the site that surfaced this, four such tags — | |
| 1258 | + * and each one stays render-blocking with no plugin lever. Unused CSS | |
| 1259 | + * skips cross-origin hrefs by design, so nothing else picks them up. | |
| 1260 | + * | |
| 1261 | + * Runs on the finished page buffer, so the rewrite is baked into the | |
| 1262 | + * cached HTML and replays on every static hit. Deliberately narrow: only | |
| 1263 | + * links whose host is on the font-CSS allowlist are touched — see | |
| 1264 | + * FONT_CSS_HOSTS. Same-origin links (no host, or the site's own) never | |
| 1265 | + * match the allowlist and are untouched. | |
| 1266 | + * | |
| 1267 | + * @param string $html Complete page HTML. | |
| 1268 | + */ | |
| 1269 | + public static function async_raw_font_css_links( $html ): string { | |
| 1270 | + if ( ! is_string( $html ) || '' === $html ) { | |
| 1271 | + return (string) $html; | |
| 1272 | + } | |
| 1273 | + if ( self::skip_in_non_frontend_context() ) { | |
| 1274 | + return $html; | |
| 1275 | + } | |
| 1276 | + $opts = self::opts(); | |
| 1277 | + if ( empty( $opts['async_css'] ) ) { | |
| 1278 | + return $html; | |
| 1279 | + } | |
| 1280 | + | |
| 1281 | + // Never rewrite inside a <noscript>. That block IS the no-JS | |
| 1282 | + // fallback — its <link> is a plain blocking stylesheet on purpose, | |
| 1283 | + // and async_style_tag() itself emits one for every sheet it defers. | |
| 1284 | + // Rewriting it would nest <noscript> (invalid; the parser closes the | |
| 1285 | + // outer block at the first </noscript>) and hand no-JS visitors a | |
| 1286 | + // media="print" sheet whose onload never runs: no stylesheet at all. | |
| 1287 | + // Splitting the buffer on <noscript> spans and rewriting only the | |
| 1288 | + // slices between them also makes the pass idempotent against | |
| 1289 | + // whatever an earlier pass emitted. | |
| 1290 | + $parts = preg_split( | |
| 1291 | + '#(<noscript\b[^>]*>.*?</noscript\s*>)#is', | |
| 1292 | + $html, | |
| 1293 | + -1, | |
| 1294 | + PREG_SPLIT_DELIM_CAPTURE | |
| 1295 | + ); | |
| 1296 | + | |
| 1297 | + // preg_split failed (pathological buffer / backtrack limit). Without | |
| 1298 | + // the split we cannot tell a fallback link from a live one, so leave | |
| 1299 | + // the page untouched — a few blocking font sheets beat a broken | |
| 1300 | + // no-JS fallback. | |
| 1301 | + if ( ! is_array( $parts ) ) { | |
| 1302 | + return $html; | |
| 1303 | + } | |
| 1304 | + | |
| 1305 | + foreach ( $parts as $i => $part ) { | |
| 1306 | + // Odd indices are the captured <noscript> blocks. | |
| 1307 | + if ( 1 === $i % 2 || '' === $part ) { | |
| 1308 | + continue; | |
| 1309 | + } | |
| 1310 | + $parts[ $i ] = self::async_font_links_in_slice( $part ); | |
| 1311 | + } | |
| 1312 | + | |
| 1313 | + return implode( '', $parts ); | |
| 1314 | + } | |
| 1315 | + | |
| 1316 | + /** | |
| 1317 | + * Rewrite the font-CSS links in one <noscript>-free slice of the buffer. | |
| 1318 | + * | |
| 1319 | + * @param string $html Slice of page HTML with no <noscript> spans. | |
| 1320 | + */ | |
| 1321 | + private static function async_font_links_in_slice( string $html ): string { | |
| 1322 | + $hosts = self::font_css_hosts(); | |
| 1323 | + | |
| 1324 | + $out = preg_replace_callback( | |
| 1325 | + '#<link\b[^>]*>#i', | |
| 1326 | + static function ( array $m ) use ( $hosts ): string { | |
| 1327 | + $tag = $m[0]; | |
| 1328 | + | |
| 1329 | + // Only plain stylesheets — never preload/alternate/anything | |
| 1330 | + // carrying explicit author intent. `(?<![-\w])` not `\b`, so | |
| 1331 | + // a `data-rel=` attribute can never read as the rel — same | |
| 1332 | + // reason the delay passes spell src that way. (#273) | |
| 1333 | + if ( ! preg_match( '#(?<![-\w])rel\s*=\s*(["\']?)\s*stylesheet\s*\1#i', $tag ) ) { | |
| 1334 | + return $tag; | |
| 1335 | + } | |
| 1336 | + | |
| 1337 | + // Already deferred (either marker spelling — ours and Pro's), | |
| 1338 | + // or explicitly opted out by the theme. | |
| 1339 | + foreach ( array( 'data-xs-async', 'data-xspeed-async', 'data-xspeed-keep' ) as $marker ) { | |
| 1340 | + if ( false !== stripos( $tag, $marker ) ) { | |
| 1341 | + return $tag; | |
| 1342 | + } | |
| 1343 | + } | |
| 1344 | + | |
| 1345 | + // An onload handler on a stylesheet link is only ever | |
| 1346 | + // somebody's finished async pattern — same rule as | |
| 1347 | + // async_style_tag(). (#216) | |
| 1348 | + if ( preg_match( '#(?<![-\w])onload\s*=#i', $tag ) ) { | |
| 1349 | + return $tag; | |
| 1350 | + } | |
| 1351 | + | |
| 1352 | + // A sheet that never applies on screen is not blocking paint. | |
| 1353 | + if ( preg_match( '#(?<![-\w])media\s*=\s*(["\'])([^"\']*)\1#i', $tag, $mm ) | |
| 1354 | + && in_array( strtolower( trim( $mm[2] ) ), self::NON_SCREEN_MEDIA, true ) ) { | |
| 1355 | + return $tag; | |
| 1356 | + } | |
| 1357 | + | |
| 1358 | + if ( ! preg_match( '#(?<![-\w])href\s*=\s*(["\'])([^"\']+)\1#i', $tag, $hm ) ) { | |
| 1359 | + return $tag; | |
| 1360 | + } | |
| 1361 | + // No host means a relative URL — same-origin, and the enqueue | |
| 1362 | + // path's business if it is anybody's. | |
| 1363 | + $host = strtolower( (string) wp_parse_url( $hm[2], PHP_URL_HOST ) ); | |
| 1364 | + if ( '' === $host || ! in_array( $host, $hosts, true ) ) { | |
| 1365 | + return $tag; | |
| 1366 | + } | |
| 1367 | + | |
| 1368 | + return self::async_link_markup( $tag ); | |
| 1369 | + }, | |
| 1370 | + $html | |
| 1371 | + ); | |
| 1372 | + | |
| 1373 | + // A PCRE failure returns null — the unrewritten slice is the safe | |
| 1374 | + // fallback, never an empty page. | |
| 1375 | + return null === $out ? $html : $out; | |
| 1376 | + } | |
| 1377 | + | |
| 1378 | + /** | |
| 528 | 1379 | * Whether a stylesheet handle carries the page's layout, and so must |
| 529 | 1380 | * keep blocking the first paint. |
| 530 | 1381 | * |
| 531 | 1382 | * Two families qualify: |
| @@ -624,8 +1475,27 @@ | ||
| 624 | 1475 | * |
| 625 | 1476 | * Skip URLs whose query carries non-ver params — those might be |
| 626 | 1477 | * intentional (e.g. a CDN providing per-image transforms). |
| 627 | 1478 | * |
| 1479 | + * `ver` is load-bearing on one class of asset: a file a plugin | |
| 1480 | + * REGENERATES IN PLACE. Complianz rewrites | |
| 1481 | + * uploads/complianz/css/banner-1-optin.css whenever the banner is | |
| 1482 | + * edited, Beaver Builder rewrites uploads/bb-plugin/cache/<post>-layout.css | |
| 1483 | + * on every layout save, Elementor uploads/elementor/css/post-<id>.css on | |
| 1484 | + * publish. The path never changes, so `?ver=<timestamp|hash>` is the only | |
| 1485 | + * thing telling a browser — or our own Browser Cache `immutable` rule — to | |
| 1486 | + * refetch. Strip it and the old styling is served until the browser cache | |
| 1487 | + * gives up, which for us is a year. So anything under the uploads root | |
| 1488 | + * keeps its version. | |
| 1489 | + * | |
| 1490 | + * Release assets under plugins/, themes/ and core are still stripped, but | |
| 1491 | + * not because they are safe: an update overwrites the same path there too, | |
| 1492 | + * and only `?ver=` changed. The difference is frequency, not mechanism — a | |
| 1493 | + * plugin update lands rarely and is expected to, a banner edit is a setting | |
| 1494 | + * the user just changed and expects to see. Stripping is the feature the | |
| 1495 | + * toggle is for; with Browser Cache on it is what the user is buying, and | |
| 1496 | + * `docs/user/minification.md` states the cost. (#276) | |
| 1497 | + * | |
| 628 | 1498 | * @param string $src |
| 629 | 1499 | */ |
| 630 | 1500 | public static function strip_version_query( $src ): string { |
| 631 | 1501 | if ( ! is_string( $src ) || '' === $src ) { |
| @@ -638,17 +1508,46 @@ | ||
| 638 | 1508 | if ( ! is_array( $parts ) || empty( $parts['query'] ) ) { |
| 639 | 1509 | return $src; |
| 640 | 1510 | } |
| 641 | 1511 | parse_str( $parts['query'], $query ); |
| 642 | - if ( ! is_array( $query ) ) { | |
| 1512 | + if ( ! is_array( $query ) || ! array_key_exists( 'ver', $query ) ) { | |
| 643 | 1513 | return $src; |
| 644 | 1514 | } |
| 1515 | + | |
| 1516 | + $strip = ! self::is_regenerated_asset( $parts ); | |
| 1517 | + | |
| 1518 | + /** | |
| 1519 | + * Whether Remove Query Strings drops `?ver` from this asset URL. | |
| 1520 | + * | |
| 1521 | + * False by default under the uploads root, where page builders and | |
| 1522 | + * consent plugins rewrite generated CSS/JS in place and `ver` is its | |
| 1523 | + * only cache-buster. Return false to protect a generator that writes | |
| 1524 | + * somewhere else, true to force stripping. | |
| 1525 | + * | |
| 1526 | + * @param bool $strip Whether `ver` will be removed. | |
| 1527 | + * @param string $src The asset URL as enqueued. | |
| 1528 | + */ | |
| 1529 | + if ( ! apply_filters( 'xspeed_strip_asset_version', $strip, $src ) ) { | |
| 1530 | + return $src; | |
| 1531 | + } | |
| 1532 | + | |
| 645 | 1533 | // Only strip 'ver' — keep anything else the asset URL needs. |
| 646 | 1534 | unset( $query['ver'] ); |
| 647 | 1535 | $new_query = http_build_query( $query ); |
| 648 | - $new_url = ( $parts['scheme'] ?? 'http' ) . '://' . ( $parts['host'] ?? '' ); | |
| 649 | - if ( isset( $parts['port'] ) ) { | |
| 650 | - $new_url .= ':' . $parts['port']; | |
| 1536 | + | |
| 1537 | + // Rebuild the authority only when the source had one. An enqueued | |
| 1538 | + // src is not always absolute: `//cdn.example/x.css` says "the | |
| 1539 | + // page's own scheme", and defaulting that to http:// is mixed | |
| 1540 | + // content an https page blocks outright; `/wp-includes/x.js` has no | |
| 1541 | + // host at all, and pasting one in produced `http:///wp-includes/…`, | |
| 1542 | + // which resolves nowhere. | |
| 1543 | + $new_url = ''; | |
| 1544 | + if ( isset( $parts['host'] ) && '' !== $parts['host'] ) { | |
| 1545 | + $new_url = isset( $parts['scheme'] ) ? $parts['scheme'] . '://' : '//'; | |
| 1546 | + $new_url .= $parts['host']; | |
| 1547 | + if ( isset( $parts['port'] ) ) { | |
| 1548 | + $new_url .= ':' . $parts['port']; | |
| 1549 | + } | |
| 651 | 1550 | } |
| 652 | 1551 | $new_url .= $parts['path'] ?? ''; |
| 653 | 1552 | if ( '' !== $new_query ) { |
| 654 | 1553 | $new_url .= '?' . $new_query; |
| @@ -659,8 +1558,80 @@ | ||
| 659 | 1558 | return $new_url; |
| 660 | 1559 | } |
| 661 | 1560 | |
| 662 | 1561 | /** |
| 1562 | + * Memoised uploads root, see uploads_base(). Cleared by reset_state(). | |
| 1563 | + * | |
| 1564 | + * @var array{host:string,path:string}|null | |
| 1565 | + */ | |
| 1566 | + private static $uploads_base = null; | |
| 1567 | + | |
| 1568 | + /** | |
| 1569 | + * The uploads root as a URL host + PATH, read from wp_get_upload_dir() | |
| 1570 | + * rather than hardcoded so a moved uploads dir, the `UPLOADS` constant and | |
| 1571 | + * the legacy multisite `/files/` layout all work. | |
| 1572 | + * | |
| 1573 | + * On multisite wp_get_upload_dir() answers with the per-site | |
| 1574 | + * `…/uploads/sites/<id>`. Generated assets live under the network root | |
| 1575 | + * too, so the suffix comes off and the whole tree matches. | |
| 1576 | + * | |
| 1577 | + * @return array{host:string,path:string} | |
| 1578 | + */ | |
| 1579 | + private static function uploads_base(): array { | |
| 1580 | + if ( null !== self::$uploads_base ) { | |
| 1581 | + return self::$uploads_base; | |
| 1582 | + } | |
| 1583 | + $base = ''; | |
| 1584 | + if ( function_exists( 'wp_get_upload_dir' ) ) { | |
| 1585 | + $dir = wp_get_upload_dir(); | |
| 1586 | + $base = is_array( $dir ) && isset( $dir['baseurl'] ) ? (string) $dir['baseurl'] : ''; | |
| 1587 | + } | |
| 1588 | + $host = ''; | |
| 1589 | + $path = ''; | |
| 1590 | + if ( '' !== $base ) { | |
| 1591 | + $host = strtolower( (string) wp_parse_url( $base, PHP_URL_HOST ) ); | |
| 1592 | + $path = (string) wp_parse_url( $base, PHP_URL_PATH ); | |
| 1593 | + } | |
| 1594 | + $path = (string) preg_replace( '#/sites/\d+/?$#', '', rtrim( $path, '/' ) ); | |
| 1595 | + if ( '' === $path && '' === $host ) { | |
| 1596 | + // Unreadable. An empty prefix would match every asset on the | |
| 1597 | + // site, so fall back to where uploads normally is. | |
| 1598 | + $path = '/wp-content/uploads'; | |
| 1599 | + } | |
| 1600 | + self::$uploads_base = array( | |
| 1601 | + 'host' => $host, | |
| 1602 | + 'path' => $path, | |
| 1603 | + ); | |
| 1604 | + return self::$uploads_base; | |
| 1605 | + } | |
| 1606 | + | |
| 1607 | + /** | |
| 1608 | + * Does this URL sit under the uploads root — i.e. is it a file some plugin | |
| 1609 | + * generates at runtime and rewrites in place? | |
| 1610 | + * | |
| 1611 | + * @param array<string,mixed> $parts wp_parse_url() output for the asset. | |
| 1612 | + */ | |
| 1613 | + private static function is_regenerated_asset( array $parts ): bool { | |
| 1614 | + $base = self::uploads_base(); | |
| 1615 | + | |
| 1616 | + if ( '' !== $base['path'] ) { | |
| 1617 | + // Path only, never host: a pull-zone CDN, a protocol-relative URL | |
| 1618 | + // and an http/https flip all leave the path alone. | |
| 1619 | + $path = (string) ( $parts['path'] ?? '' ); | |
| 1620 | + return '' !== $path && 0 === strpos( $path, $base['path'] . '/' ); | |
| 1621 | + } | |
| 1622 | + | |
| 1623 | + // Uploads AT the root of their own domain — an offload plugin | |
| 1624 | + // pointing `upload_url_path` at https://cdn.example.com. There is no | |
| 1625 | + // prefix left to test, and testing the path anyway would have read | |
| 1626 | + // every generated file on that CDN as an ordinary release asset and | |
| 1627 | + // stripped the one thing telling a browser it had changed. The host | |
| 1628 | + // is the whole answer here: everything served from it is an upload. | |
| 1629 | + $host = strtolower( (string) ( $parts['host'] ?? '' ) ); | |
| 1630 | + return '' !== $host && $host === $base['host']; | |
| 1631 | + } | |
| 1632 | + | |
| 1633 | + /** | |
| 663 | 1634 | * Defensive context guard for filter callbacks. Mirrors the registration- |
| 664 | 1635 | * time bail in Minifier::__construct() so a late context flip (admin page |
| 665 | 1636 | * render kicked off mid-request, REST_REQUEST set after plugins_loaded, |
| 666 | 1637 | * etc.) doesn't let frontend tag rewrites leak into wp-admin / AJAX / |
| @@ -703,8 +1674,19 @@ | ||
| 703 | 1674 | private static function is_excluded_script( string $handle, string $src ): bool { |
| 704 | 1675 | if ( in_array( $handle, self::ALWAYS_EXCLUDED_HANDLES, true ) ) { |
| 705 | 1676 | return true; |
| 706 | 1677 | } |
| 1678 | + // Never defer or delay our own scripts. The fold and RUM beacons | |
| 1679 | + // measure the FIRST paint — delayed to first interaction they | |
| 1680 | + // measure a scrolled page or nothing, so fold quorum never fills | |
| 1681 | + // and full CSS deferral never licenses. Found live: delay_js with | |
| 1682 | + // empty targets delayed the fold beacon itself, and the site sat | |
| 1683 | + // at zero fold reports for hours while its stylesheets stayed | |
| 1684 | + // render-blocking. Prefix, not a handle list, so a Pro module's | |
| 1685 | + // beacon added later cannot re-open the hole. | |
| 1686 | + if ( 0 === strpos( $handle, 'xspeed-' ) ) { | |
| 1687 | + return true; | |
| 1688 | + } | |
| 707 | 1689 | $opts = self::opts(); |
| 708 | 1690 | $excluded = is_array( $opts['defer_js_excluded'] ?? null ) ? $opts['defer_js_excluded'] : array(); |
| 709 | 1691 | if ( empty( $excluded ) ) { |
| 710 | 1692 | return false; |
| @@ -727,8 +1709,52 @@ | ||
| 727 | 1709 | * page's JS. Empty targets = historical behavior (delay everything |
| 728 | 1710 | * minus exclusions). Same matching semantics as the exclusion list: |
| 729 | 1711 | * exact handle match OR case-insensitive URL substring. |
| 730 | 1712 | */ |
| 1713 | + /** | |
| 1714 | + * Whether the user's delay_js_targets list matches this haystack. | |
| 1715 | + * | |
| 1716 | + * The inline-snippet pass needs the target list WITHOUT | |
| 1717 | + * is_delay_target()'s empty-list-means-everything default — an inline | |
| 1718 | + * body is only ever delayed on a positive match. | |
| 1719 | + * | |
| 1720 | + * @param string $haystack Script body (or URL) to match fragments against. | |
| 1721 | + */ | |
| 1722 | + private static function matches_user_targets( string $haystack ): bool { | |
| 1723 | + $opts = self::opts(); | |
| 1724 | + $targets = is_array( $opts['delay_js_targets'] ?? null ) ? $opts['delay_js_targets'] : array(); | |
| 1725 | + foreach ( $targets as $needle ) { | |
| 1726 | + $needle = (string) $needle; | |
| 1727 | + if ( '' !== $needle && false !== stripos( $haystack, $needle ) ) { | |
| 1728 | + return true; | |
| 1729 | + } | |
| 1730 | + } | |
| 1731 | + return false; | |
| 1732 | + } | |
| 1733 | + | |
| 1734 | + /** | |
| 1735 | + * Whether the user EXPLICITLY named this script in delay_js_targets. | |
| 1736 | + * | |
| 1737 | + * Unlike is_delay_target() this never treats an empty list as | |
| 1738 | + * everything and never falls back to the vendor list — it answers | |
| 1739 | + * only "did the user deliberately point at this handle/URL?", which | |
| 1740 | + * is what lets an explicit entry override the inline-bound guard. | |
| 1741 | + * | |
| 1742 | + * @param string $handle Script handle. | |
| 1743 | + * @param string $src Script URL. | |
| 1744 | + */ | |
| 1745 | + private static function is_user_named_target( string $handle, string $src ): bool { | |
| 1746 | + $opts = self::opts(); | |
| 1747 | + $targets = is_array( $opts['delay_js_targets'] ?? null ) ? $opts['delay_js_targets'] : array(); | |
| 1748 | + foreach ( $targets as $needle ) { | |
| 1749 | + $needle = (string) $needle; | |
| 1750 | + if ( '' !== $needle && self::target_matches( $needle, $handle, $src ) ) { | |
| 1751 | + return true; | |
| 1752 | + } | |
| 1753 | + } | |
| 1754 | + return false; | |
| 1755 | + } | |
| 1756 | + | |
| 731 | 1757 | private static function is_delay_target( string $handle, string $src ): bool { |
| 732 | 1758 | $opts = self::opts(); |
| 733 | 1759 | $targets = is_array( $opts['delay_js_targets'] ?? null ) ? $opts['delay_js_targets'] : array(); |
| 734 | 1760 | $targets = array_filter( array_map( 'strval', $targets ), static fn( $t ) => '' !== $t ); |
| @@ -739,8 +1765,116 @@ | ||
| 739 | 1765 | if ( self::target_matches( $needle, $handle, $src ) ) { |
| 740 | 1766 | return true; |
| 741 | 1767 | } |
| 742 | 1768 | } |
| 1769 | + // The user's list is an ALLOW-list, so a target they never thought to | |
| 1770 | + // add is not delayed — and the scripts worth delaying are third-party | |
| 1771 | + // tags nobody enumerates by hand. Falling back to the built-in vendor | |
| 1772 | + // list means a site that lists one heavy embed still gets the obvious | |
| 1773 | + // analytics and widget tags postponed, instead of silently keeping | |
| 1774 | + // them on the main thread. (A user who wants one of these to run | |
| 1775 | + // early excludes it; the exclusion list is checked before this.) | |
| 1776 | + return self::matches_known_third_party( $src ); | |
| 1777 | + } | |
| 1778 | + | |
| 1779 | + /** | |
| 1780 | + * Whether a URL belongs to a third-party tag that is safe to postpone. | |
| 1781 | + * | |
| 1782 | + * These are analytics, tag managers, chat widgets, review embeds, session | |
| 1783 | + * recorders and error trackers: scripts that never paint anything above | |
| 1784 | + * the fold and that no first-party code holds a synchronous reference to. | |
| 1785 | + * They are also the scripts that dominate a real page's blocking time — | |
| 1786 | + * on embedpress.com one chat widget alone accounted for ~450ms of TBT and | |
| 1787 | + * a 22-point score swing between runs, purely on whether it happened to | |
| 1788 | + * arrive inside the measurement window. | |
| 1789 | + * | |
| 1790 | + * Matched on URL only, never on handle: these tags are printed straight | |
| 1791 | + * into wp_head / wp_footer by their vendors' snippets and usually have no | |
| 1792 | + * WordPress handle at all. Host fragments rather than whole domains, so a | |
| 1793 | + * regional or versioned CDN path still matches. | |
| 1794 | + * | |
| 1795 | + * Deliberately NOT here: anything from the site's own origin, jQuery, or | |
| 1796 | + * any wp-* core script. Those carry inline consumers, and delaying them | |
| 1797 | + * is what breaks pages — see inline_bound_handles(). | |
| 1798 | + */ | |
| 1799 | + private const KNOWN_THIRD_PARTY_SRC = array( | |
| 1800 | + // Tag managers and analytics. | |
| 1801 | + 'googletagmanager.com', | |
| 1802 | + 'google-analytics.com', | |
| 1803 | + 'analytics.google.com', | |
| 1804 | + '/gtag/js', | |
| 1805 | + 'gtm4wp', | |
| 1806 | + 'plausible.io', | |
| 1807 | + 'matomo', | |
| 1808 | + 'segment.com/analytics.js', | |
| 1809 | + 'stats.wp.com', | |
| 1810 | + // Advertising and conversion pixels. | |
| 1811 | + 'connect.facebook.net', | |
| 1812 | + 'fbevents.js', | |
| 1813 | + 'ads-twitter.com', | |
| 1814 | + 'snap.licdn.com', | |
| 1815 | + 'analytics.tiktok.com', | |
| 1816 | + 'googleadservices.com', | |
| 1817 | + 'doubleclick.net', | |
| 1818 | + // Session recording and heatmaps. | |
| 1819 | + 'hotjar.com', | |
| 1820 | + 'clarity.ms', | |
| 1821 | + 'mouseflow.com', | |
| 1822 | + 'fullstory.com', | |
| 1823 | + 'luckyorange', | |
| 1824 | + // Chat and support widgets. | |
| 1825 | + 'client.crisp.chat', | |
| 1826 | + 'widget.intercom.io', | |
| 1827 | + 'js.driftt.com', | |
| 1828 | + 'tawk.to', | |
| 1829 | + 'livechatinc.com', | |
| 1830 | + 'zdassets.com', | |
| 1831 | + 'helpscout.net', | |
| 1832 | + // Reviews, social proof and marketing. | |
| 1833 | + 'tp.widget.bootstrap', | |
| 1834 | + 'trustpilot.com', | |
| 1835 | + 'static.klaviyo.com', | |
| 1836 | + 'js.hs-scripts.com', | |
| 1837 | + 'list-manage.com', | |
| 1838 | + 'sumo.com', | |
| 1839 | + // Error and performance monitoring. | |
| 1840 | + 'sentry-cdn.com', | |
| 1841 | + 'browser.sentry', | |
| 1842 | + 'bugsnag.com', | |
| 1843 | + 'newrelic.com', | |
| 1844 | + ); | |
| 1845 | + | |
| 1846 | + /** | |
| 1847 | + * Match a script URL against the built-in third-party list. | |
| 1848 | + * | |
| 1849 | + * @param string $src Script source URL. | |
| 1850 | + */ | |
| 1851 | + private static function matches_known_third_party( string $src ): bool { | |
| 1852 | + if ( '' === $src ) { | |
| 1853 | + return false; | |
| 1854 | + } | |
| 1855 | + | |
| 1856 | + $known = self::KNOWN_THIRD_PARTY_SRC; | |
| 1857 | + | |
| 1858 | + /** | |
| 1859 | + * URL fragments the delay pass treats as safe-to-postpone third-party | |
| 1860 | + * tags when the user's target list does not match. | |
| 1861 | + * | |
| 1862 | + * Append a vendor this list does not know yet, or remove one the site | |
| 1863 | + * genuinely needs early. Entries are case-insensitive substrings of | |
| 1864 | + * the script URL. | |
| 1865 | + * | |
| 1866 | + * @param string[] $known Built-in fragments. | |
| 1867 | + * @param string $src The script URL being tested. | |
| 1868 | + */ | |
| 1869 | + $known = (array) apply_filters( 'xspeed_delay_known_third_party', $known, $src ); | |
| 1870 | + | |
| 1871 | + foreach ( $known as $needle ) { | |
| 1872 | + $needle = (string) $needle; | |
| 1873 | + if ( '' !== $needle && false !== stripos( $src, $needle ) ) { | |
| 1874 | + return true; | |
| 1875 | + } | |
| 1876 | + } | |
| 743 | 1877 | return false; |
| 744 | 1878 | } |
| 745 | 1879 | |
| 746 | 1880 | private static function opts(): array { |
| @@ -754,10 +1888,158 @@ | ||
| 754 | 1888 | * Test-only — clear cached opts + bootstrap-printed flag. |
| 755 | 1889 | */ |
| 756 | 1890 | public static function reset_state(): void { |
| 757 | 1891 | self::$opts = null; |
| 1892 | + self::$uploads_base = null; | |
| 758 | 1893 | self::$delay_bootstrap_printed = false; |
| 759 | 1894 | self::$js_measured_layout = null; |
| 1895 | + self::$inline_bound_handles = null; | |
| 1896 | + self::$pristine_tag = array(); | |
| 1897 | + } | |
| 1898 | + | |
| 1899 | + /** | |
| 1900 | + * Per-request memo for inline_bound_handles(). Null = not resolved. | |
| 1901 | + * | |
| 1902 | + * @var array<string,true>|null | |
| 1903 | + */ | |
| 1904 | + private static $inline_bound_handles = null; | |
| 1905 | + | |
| 1906 | + /** | |
| 1907 | + * Handles that cannot be deferred because inline code depends on them. | |
| 1908 | + * | |
| 1909 | + * #234 fixed the case where a handle carries its OWN inline block: the | |
| 1910 | + * tag WordPress hands the filter is `before_inline + external + | |
| 1911 | + * after_inline`, so defer goes on the external <script> and order holds. | |
| 1912 | + * That leaves the cross-handle case, which is the one that actually | |
| 1913 | + * breaks sites: `wp_add_inline_script( 'foo', … )` prints a bare inline | |
| 1914 | + * block that runs at parse time and calls into whatever `foo` — or any | |
| 1915 | + * of foo's DEPENDENCIES — defined. Inline scripts can never be deferred | |
| 1916 | + * (the HTML spec ignores the attribute), so deferring anything they read | |
| 1917 | + * from inverts the order WordPress guarantees and throws on a global | |
| 1918 | + * that is not there yet. | |
| 1919 | + * | |
| 1920 | + * jQuery is the canonical victim: one `wp_add_inline_script( 'jquery', | |
| 1921 | + * 'jQuery(function($){…})' )` anywhere on the page makes `jquery-core` | |
| 1922 | + * undeferrable, and every hand-maintained exclusion list in the wild | |
| 1923 | + * exists to say so. The registry already knows it, so read it instead of | |
| 1924 | + * asking the user. | |
| 1925 | + * | |
| 1926 | + * Walks each handle carrying `after`/`before` inline data and marks the | |
| 1927 | + * handle plus its transitive dependency chain. Cycles are guarded by the | |
| 1928 | + * seen-map, so a self- or mutually-referential deps array terminates. | |
| 1929 | + * | |
| 1930 | + * Pure aside from the global registry read; memoised per request and | |
| 1931 | + * cleared by reset_state(). | |
| 1932 | + * | |
| 1933 | + * @return array<string,true> Handle => true, for O(1) lookup. | |
| 1934 | + */ | |
| 1935 | + public static function inline_bound_handles(): array { | |
| 1936 | + if ( null !== self::$inline_bound_handles ) { | |
| 1937 | + return self::$inline_bound_handles; | |
| 1938 | + } | |
| 1939 | + | |
| 1940 | + $bound = array(); | |
| 1941 | + if ( function_exists( 'wp_scripts' ) ) { | |
| 1942 | + $scripts = wp_scripts(); | |
| 1943 | + if ( $scripts instanceof \WP_Scripts ) { | |
| 1944 | + foreach ( array_keys( (array) $scripts->registered ) as $handle ) { | |
| 1945 | + $handle = (string) $handle; | |
| 1946 | + if ( ! self::handle_carries_inline( $scripts, $handle ) ) { | |
| 1947 | + continue; | |
| 1948 | + } | |
| 1949 | + self::mark_with_deps( $scripts, $handle, $bound ); | |
| 1950 | + } | |
| 1951 | + } | |
| 1952 | + } | |
| 1953 | + | |
| 1954 | + /** | |
| 1955 | + * Handles auto-excluded from defer because inline code reads them. | |
| 1956 | + * | |
| 1957 | + * Return a handle => true map. Add an entry to protect a script whose | |
| 1958 | + * inline consumer this cannot see (one printed directly by a theme | |
| 1959 | + * rather than through wp_add_inline_script), or remove one to defer a | |
| 1960 | + * handle whose inline block is known not to touch it. | |
| 1961 | + * | |
| 1962 | + * @param array<string,true> $bound Detected handles. | |
| 1963 | + */ | |
| 1964 | + $bound = (array) apply_filters( 'xspeed_defer_inline_bound_handles', $bound ); | |
| 1965 | + | |
| 1966 | + self::$inline_bound_handles = $bound; | |
| 1967 | + | |
| 1968 | + return self::$inline_bound_handles; | |
| 1969 | + } | |
| 1970 | + | |
| 1971 | + /** | |
| 1972 | + * Whether a handle must be kept out of a combined bundle. | |
| 1973 | + * | |
| 1974 | + * Combining re-homes a script's code under a different handle, so every | |
| 1975 | + * protection keyed to the ORIGINAL handle or URL stops matching: the | |
| 1976 | + * user's `defer_js_excluded` entry, and the inline-bound set above. The | |
| 1977 | + * combiner already refuses a handle carrying its own inline data, which | |
| 1978 | + * is why the gap is invisible until you look for it — a DEPENDENCY of an | |
| 1979 | + * inline consumer carries none of its own, so `jquery-core` lands in the | |
| 1980 | + * bundle while the exclusion list still reads as though it were honoured. | |
| 1981 | + * | |
| 1982 | + * Returning true here is enough on its own: the combiner drops any | |
| 1983 | + * dependent of an uncombinable handle transitively, so the whole chain | |
| 1984 | + * stays in the queue where WordPress prints it in the right order. | |
| 1985 | + * | |
| 1986 | + * @param string $handle Script handle. | |
| 1987 | + * @param string $src Registered source URL. | |
| 1988 | + */ | |
| 1989 | + public static function is_protected_from_bundling( string $handle, string $src ): bool { | |
| 1990 | + if ( self::is_excluded_script( $handle, $src ) ) { | |
| 1991 | + return true; | |
| 1992 | + } | |
| 1993 | + return isset( self::inline_bound_handles()[ $handle ] ); | |
| 1994 | + } | |
| 1995 | + | |
| 1996 | + /** | |
| 1997 | + * Whether a handle has inline JS attached in either position. | |
| 1998 | + * | |
| 1999 | + * `get_data()` returns the raw value, which is an array of code chunks | |
| 2000 | + * for `after` and a string for `before`; both are falsy when absent, and | |
| 2001 | + * an empty chunk array must not count as inline code. | |
| 2002 | + * | |
| 2003 | + * @param \WP_Scripts $scripts Registry. | |
| 2004 | + * @param string $handle Handle to inspect. | |
| 2005 | + */ | |
| 2006 | + private static function handle_carries_inline( \WP_Scripts $scripts, string $handle ): bool { | |
| 2007 | + foreach ( array( 'after', 'before' ) as $position ) { | |
| 2008 | + $data = $scripts->get_data( $handle, $position ); | |
| 2009 | + if ( is_array( $data ) ) { | |
| 2010 | + foreach ( $data as $chunk ) { | |
| 2011 | + if ( '' !== trim( (string) $chunk ) ) { | |
| 2012 | + return true; | |
| 2013 | + } | |
| 2014 | + } | |
| 2015 | + continue; | |
| 2016 | + } | |
| 2017 | + if ( '' !== trim( (string) $data ) ) { | |
| 2018 | + return true; | |
| 2019 | + } | |
| 2020 | + } | |
| 2021 | + return false; | |
| 2022 | + } | |
| 2023 | + | |
| 2024 | + /** | |
| 2025 | + * Mark a handle and everything it depends on, transitively. | |
| 2026 | + * | |
| 2027 | + * @param \WP_Scripts $scripts Registry. | |
| 2028 | + * @param string $handle Handle to mark. | |
| 2029 | + * @param array<string,true> $seen Accumulator, by reference. | |
| 2030 | + */ | |
| 2031 | + private static function mark_with_deps( \WP_Scripts $scripts, string $handle, array &$seen ): void { | |
| 2032 | + if ( isset( $seen[ $handle ] ) ) { | |
| 2033 | + return; | |
| 2034 | + } | |
| 2035 | + $seen[ $handle ] = true; | |
| 2036 | + if ( ! isset( $scripts->registered[ $handle ]->deps ) ) { | |
| 2037 | + return; | |
| 2038 | + } | |
| 2039 | + foreach ( (array) $scripts->registered[ $handle ]->deps as $dep ) { | |
| 2040 | + self::mark_with_deps( $scripts, (string) $dep, $seen ); | |
| 2041 | + } | |
| 760 | 2042 | } |
| 761 | 2043 | |
| 762 | 2044 | /** |
| 763 | 2045 | * Per-request memo for page_has_js_measured_layout(). Null = not resolved. |
