| @@ -148,26 +148,37 @@ | ||
| 148 | 148 | return $out; |
| 149 | 149 | } |
| 150 | 150 | |
| 151 | 151 | /** |
| 152 | - * Strip core `loading="lazy"` and add `fetchpriority="high"` + | |
| 153 | - * `decoding="async"` on every <img> whose tag matches one of the given | |
| 154 | - * exclusion substrings. Mirrors what Lazy_Loader does for an excluded image | |
| 155 | - * inside the_content, but page-wide so heroes outside it are covered too. | |
| 156 | - * (FBS-83553 H2) | |
| 152 | + * Strip core `loading="lazy"` and add `decoding="async"` on every <img> | |
| 153 | + * whose tag matches one of the given exclusion substrings. Mirrors what | |
| 154 | + * Lazy_Loader does for an excluded image inside the_content, but page-wide | |
| 155 | + * so heroes outside it are covered too. (FBS-83553 H2) | |
| 157 | 156 | * |
| 157 | + * Only the first visible match outside <footer>/<nav>/<aside> gets | |
| 158 | + * `fetchpriority="high"`, and only when no image in the page holds it yet. | |
| 159 | + * Every match used to get it, so a pattern naming the header and footer | |
| 160 | + * logo, or a hero in a closed accordion, put several images at High. (#558) | |
| 161 | + * | |
| 158 | 162 | * @param string $html Full page HTML. |
| 159 | 163 | * @param string[] $exclusions Substring patterns identifying above-the-fold heroes. |
| 160 | 164 | */ |
| 161 | 165 | private static function promote_excluded_images( string $html, array $exclusions ): string { |
| 162 | - return (string) preg_replace_callback( | |
| 163 | - '#<img\b(?:"[^"]*"|\'[^\']*\'|[^>"\'])*>#i', | |
| 164 | - static function ( array $m ) use ( $exclusions ) { | |
| 165 | - $tag = $m[0]; | |
| 166 | + $img_re = '#<img\b(?:"[^"]*"|\'[^\']*\'|[^>"\'])*>#i'; | |
| 167 | + $claimed = (bool) preg_match( '#<img\b(?:"[^"]*"|\'[^\']*\'|[^>"\'])*?(?<![-\w])fetchpriority\s*=\s*["\']?high\b#i', $html ); | |
| 168 | + $skip = $claimed ? array() : array_merge( self::chrome_container_ranges( $html ), Lazy_Loader::hidden_ranges( $html ) ); | |
| 169 | + | |
| 170 | + $result = preg_replace_callback( | |
| 171 | + $img_re, | |
| 172 | + static function ( array $m ) use ( $exclusions, &$claimed, $skip ) { | |
| 173 | + [ $tag, $offset ] = $m[0]; | |
| 166 | 174 | foreach ( $exclusions as $needle ) { |
| 167 | 175 | if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { |
| 168 | 176 | $tag = (string) preg_replace( '#\s*\bloading=(["\'])\s*lazy\s*\1#i', '', $tag ); |
| 169 | - $tag = self::set_fetchpriority( $tag ); | |
| 177 | + if ( ! $claimed && ! self::offset_in_ranges( (int) $offset, $skip ) && ! Lazy_Loader::tag_is_hidden( $tag, 'img' ) ) { | |
| 178 | + $tag = self::set_fetchpriority( $tag ); | |
| 179 | + $claimed = true; | |
| 180 | + } | |
| 170 | 181 | if ( ! preg_match( '#\bdecoding=#i', $tag ) ) { |
| 171 | 182 | $tag = (string) preg_replace( '#<img\b#i', '<img decoding="async"', $tag, 1 ); |
| 172 | 183 | } |
| 173 | 184 | return $tag; |
| @@ -174,10 +185,15 @@ | ||
| 174 | 185 | } |
| 175 | 186 | } |
| 176 | 187 | return $tag; |
| 177 | 188 | }, |
| 178 | - $html | |
| 189 | + $html, | |
| 190 | + -1, | |
| 191 | + $count, | |
| 192 | + PREG_OFFSET_CAPTURE | |
| 179 | 193 | ); |
| 194 | + | |
| 195 | + return is_string( $result ) ? $result : $html; | |
| 180 | 196 | } |
| 181 | 197 | |
| 182 | 198 | /** |
| 183 | 199 | * Build preconnect <link>s for detected font hosts + user hosts. |
| @@ -256,22 +272,21 @@ | ||
| 256 | 272 | // one that matters, and the feature reported success either way. The |
| 257 | 273 | // marker list and size gate were heuristics layered on top of the |
| 258 | 274 | // wrong primitive rather than replacing it. (#96) |
| 259 | 275 | $candidates = array(); |
| 260 | - $skip_ranges = self::chrome_container_ranges( $html ); | |
| 276 | + // Markup hidden on arrival (a closed <details>, `hidden`, inline | |
| 277 | + // display:none) cannot paint as LCP either. Preloading the hero of a | |
| 278 | + // closed accordion spent the page's one High fetch on it. (#558) | |
| 279 | + $skip_ranges = array_merge( self::chrome_container_ranges( $html ), Lazy_Loader::hidden_ranges( $html ) ); | |
| 280 | + // <img> indexes that can't be the LCP however they are marked: chrome, | |
| 281 | + // hidden, or a logo/icon. Only these give up a stray `high`. (#558) | |
| 282 | + $ruled_out = array(); | |
| 261 | 283 | if ( preg_match_all( '#<img\b[^>]*>#i', $html, $matches, PREG_OFFSET_CAPTURE ) ) { |
| 262 | 284 | foreach ( $matches[0] as $index => $match ) { |
| 263 | 285 | [ $tag, $offset ] = $match; |
| 264 | 286 | |
| 265 | 287 | // Skip anything the user excluded. |
| 266 | - $excluded = false; | |
| 267 | - foreach ( $exclusions as $needle ) { | |
| 268 | - if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 269 | - $excluded = true; | |
| 270 | - break; | |
| 271 | - } | |
| 272 | - } | |
| 273 | - if ( $excluded ) { | |
| 288 | + if ( self::matches_any( $tag, $exclusions ) ) { | |
| 274 | 289 | continue; |
| 275 | 290 | } |
| 276 | 291 | |
| 277 | 292 | // An image inside <footer>/<nav>/<aside> is site chrome by |
| @@ -278,9 +293,10 @@ | ||
| 278 | 293 | // construction — a footer brand strip or FAQ illustration can |
| 279 | 294 | // never be the LCP element, whatever size it declares. On the |
| 280 | 295 | // FBS-84576 repro these decoys outranked the real hero three |
| 281 | 296 | // times on one layout. |
| 282 | - if ( self::offset_in_ranges( $offset, $skip_ranges ) ) { | |
| 297 | + if ( self::offset_in_ranges( $offset, $skip_ranges ) || Lazy_Loader::tag_is_hidden( $tag, 'img' ) ) { | |
| 298 | + $ruled_out[ $index ] = true; | |
| 283 | 299 | continue; |
| 284 | 300 | } |
| 285 | 301 | |
| 286 | 302 | // An image the theme explicitly lazy-loads is never the |
| @@ -302,8 +318,9 @@ | ||
| 302 | 318 | |
| 303 | 319 | // Chrome markers / explicit opt-out / obviously-tiny images |
| 304 | 320 | // never compete. (FBS-83553 H1 "logo before hero".) |
| 305 | 321 | if ( self::looks_too_small( $tag ) ) { |
| 322 | + $ruled_out[ $index ] = true; | |
| 306 | 323 | continue; |
| 307 | 324 | } |
| 308 | 325 | |
| 309 | 326 | $candidates[] = array( |
| @@ -312,8 +329,9 @@ | ||
| 312 | 329 | 'srcset' => $srcset, |
| 313 | 330 | 'sizes' => $sizes, |
| 314 | 331 | 'score' => self::weighted_score( self::lcp_score( $tag, $srcset ), $tag, $index ), |
| 315 | 332 | 'order' => $index, |
| 333 | + 'offset' => $offset, | |
| 316 | 334 | ); |
| 317 | 335 | } |
| 318 | 336 | } |
| 319 | 337 | |
| @@ -328,9 +346,47 @@ | ||
| 328 | 346 | foreach ( self::background_candidates( $html, $exclusions, $skip_ranges ) as $bg ) { |
| 329 | 347 | $candidates[] = $bg; |
| 330 | 348 | } |
| 331 | 349 | |
| 350 | + // PASS 1c — <video poster="…">. A full-screen hero video paints its | |
| 351 | + // poster first, and that first frame IS the LCP; measured on a live | |
| 352 | + // page, a preloaded poster cut the LCP load delay from 1.5 s to 21 ms. | |
| 353 | + foreach ( self::video_poster_candidates( $html, $exclusions, $skip_ranges ) as $vp ) { | |
| 354 | + $candidates[] = $vp; | |
| 355 | + } | |
| 356 | + | |
| 357 | + // PASS 1d — background rules in inline <style> blocks. Page builders | |
| 358 | + // put the hero's background-image in generated per-post CSS printed | |
| 359 | + // inline (Elementor's `.elementor-N .elementor-element-X` rules), not | |
| 360 | + // in a style attribute — so PASS 1b never saw the element that | |
| 361 | + // actually paints as LCP on five of seven measured sites. External | |
| 362 | + // stylesheets stay out for the same reasons as before (#247): fetching | |
| 363 | + // CSS from an output-buffer pass costs more than the preload saves. | |
| 364 | + foreach ( self::style_block_candidates( $html, $exclusions, $skip_ranges ) as $sb ) { | |
| 365 | + $candidates[] = $sb; | |
| 366 | + } | |
| 367 | + | |
| 368 | + // A background video with no poster, ahead of every candidate, is the | |
| 369 | + // hero. Nothing in its box is preloadable, and every image after it | |
| 370 | + // sits lower on the page. Preloading the first of those spent the one | |
| 371 | + // high-priority fetch on an image below the fold, ahead of the CSS. | |
| 372 | + // Those images also give up a stray `high` from the lazy pass. | |
| 373 | + $video_at = self::background_video_offset( $html, $skip_ranges ); | |
| 374 | + if ( null !== $video_at ) { | |
| 375 | + foreach ( $candidates as $i => $c ) { | |
| 376 | + if ( $c['offset'] > $video_at ) { | |
| 377 | + if ( empty( $c['background'] ) ) { | |
| 378 | + $ruled_out[ $c['order'] ] = true; | |
| 379 | + } | |
| 380 | + unset( $candidates[ $i ] ); | |
| 381 | + } | |
| 382 | + } | |
| 383 | + } | |
| 384 | + | |
| 332 | 385 | if ( empty( $candidates ) ) { |
| 386 | + if ( null !== $video_at ) { | |
| 387 | + $html = self::demote_images( $html, $ruled_out ); | |
| 388 | + } | |
| 333 | 389 | return array( $html, '' ); |
| 334 | 390 | } |
| 335 | 391 | |
| 336 | 392 | // Rank by score, biggest first. Document order breaks ties, so two |
| @@ -374,14 +430,26 @@ | ||
| 374 | 430 | // Rewrite only the winning tags. Counting occurrences rather than |
| 375 | 431 | // matching on tag text, because the same markup can legitimately |
| 376 | 432 | // appear more than once on a page and only the ranked instance should |
| 377 | 433 | // be promoted. |
| 378 | - $seen = -1; | |
| 379 | - $html = preg_replace_callback( | |
| 434 | + // | |
| 435 | + // When an <img> wins, it is the page's one High image, so an image | |
| 436 | + // that can't be the LCP (a logo, an icon, a hidden panel, chrome) | |
| 437 | + // gives up any `high` it carries. That is usually the lazy pass handing | |
| 438 | + // its slot to the first image in the_content. An image that merely scored | |
| 439 | + // lower, or that the user kept out of the pick, keeps its hint: the | |
| 440 | + // ranking can be wrong, and core or the theme may have named the real | |
| 441 | + // hero. (#558) | |
| 442 | + $demote = ! empty( $chosen ) || null !== $video_at ? $ruled_out : array(); | |
| 443 | + $seen = -1; | |
| 444 | + $html = preg_replace_callback( | |
| 380 | 445 | '#<img\b[^>]*>#i', |
| 381 | - static function ( array $m ) use ( &$seen, $chosen ) { | |
| 446 | + static function ( array $m ) use ( &$seen, $chosen, $demote ) { | |
| 382 | 447 | ++$seen; |
| 383 | 448 | if ( ! isset( $chosen[ $seen ] ) ) { |
| 449 | + if ( isset( $demote[ $seen ] ) && 'high' === strtolower( self::attr( $m[0], 'fetchpriority' ) ) ) { | |
| 450 | + return (string) preg_replace( '#\s*(?<![-\w])fetchpriority\s*=\s*(["\']?)high\1#i', '', $m[0], 1 ); | |
| 451 | + } | |
| 384 | 452 | return $m[0]; |
| 385 | 453 | } |
| 386 | 454 | // Add fetchpriority="high" AND remove any loading="lazy" the |
| 387 | 455 | // theme / WP core left on the LCP image. fetchpriority="high" |
| @@ -396,8 +464,56 @@ | ||
| 396 | 464 | return array( (string) $html, $preload ); |
| 397 | 465 | } |
| 398 | 466 | |
| 399 | 467 | /** |
| 468 | + * Byte offset of the first visible background video with no poster, or | |
| 469 | + * null when the page has none. | |
| 470 | + * | |
| 471 | + * @param string $html Full page HTML. | |
| 472 | + * @param array<int,array{0:int,1:int}> $skip_ranges Chrome and hidden spans. | |
| 473 | + */ | |
| 474 | + private static function background_video_offset( string $html, array $skip_ranges ): ?int { | |
| 475 | + $body = stripos( $html, '<body' ); | |
| 476 | + if ( ! preg_match_all( '#<video\b[^>]*>#i', $html, $m, PREG_OFFSET_CAPTURE, false === $body ? 0 : $body ) ) { | |
| 477 | + return null; | |
| 478 | + } | |
| 479 | + foreach ( $m[0] as [ $tag, $offset ] ) { | |
| 480 | + if ( self::offset_in_ranges( $offset, $skip_ranges ) || Lazy_Loader::tag_is_hidden( $tag, 'video' ) ) { | |
| 481 | + continue; | |
| 482 | + } | |
| 483 | + if ( Lazy_Loader::is_background_video_without_poster( $tag ) ) { | |
| 484 | + return $offset; | |
| 485 | + } | |
| 486 | + } | |
| 487 | + return null; | |
| 488 | + } | |
| 489 | + | |
| 490 | + /** | |
| 491 | + * Strip fetchpriority="high" from the <img> tags at the given indexes. | |
| 492 | + * | |
| 493 | + * @param string $html Page HTML. | |
| 494 | + * @param array<int,bool> $indexes <img> indexes, in document order. | |
| 495 | + */ | |
| 496 | + private static function demote_images( string $html, array $indexes ): string { | |
| 497 | + if ( empty( $indexes ) ) { | |
| 498 | + return $html; | |
| 499 | + } | |
| 500 | + $seen = -1; | |
| 501 | + $out = preg_replace_callback( | |
| 502 | + '#<img\b[^>]*>#i', | |
| 503 | + static function ( array $m ) use ( &$seen, $indexes ) { | |
| 504 | + ++$seen; | |
| 505 | + if ( isset( $indexes[ $seen ] ) && 'high' === strtolower( self::attr( $m[0], 'fetchpriority' ) ) ) { | |
| 506 | + return (string) preg_replace( '#\s*(?<![-\w])fetchpriority\s*=\s*(["\']?)high\1#i', '', $m[0], 1 ); | |
| 507 | + } | |
| 508 | + return $m[0]; | |
| 509 | + }, | |
| 510 | + $html | |
| 511 | + ); | |
| 512 | + return null === $out ? $html : $out; | |
| 513 | + } | |
| 514 | + | |
| 515 | + /** | |
| 400 | 516 | * Collect CSS `background-image` heroes as LCP candidates. |
| 401 | 517 | * |
| 402 | 518 | * Only INLINE `style` attributes are read. A background declared in an |
| 403 | 519 | * external stylesheet is invisible here by design: resolving it would mean |
| @@ -469,8 +585,9 @@ | ||
| 469 | 585 | 'score' => (float) $area, |
| 470 | 586 | // Offset so a background never ties ahead of an <img> that |
| 471 | 587 | // appeared earlier in the document; ties still break on order. |
| 472 | 588 | 'order' => 100000 + $index, |
| 589 | + 'offset' => $offset, | |
| 473 | 590 | 'background' => true, |
| 474 | 591 | ); |
| 475 | 592 | } |
| 476 | 593 | |
| @@ -477,8 +594,249 @@ | ||
| 477 | 594 | return $found; |
| 478 | 595 | } |
| 479 | 596 | |
| 480 | 597 | /** |
| 598 | + * Collect `<video poster="…">` first frames as LCP candidates. | |
| 599 | + * | |
| 600 | + * The poster is what the viewer sees until (and unless) the video plays — | |
| 601 | + * on a background-video hero, delayed by the Lazy module, it is the ONLY | |
| 602 | + * frame the initial paint has. Scored like a background: the element's | |
| 603 | + * declared inline-style area, or the unknown-size floor, with the order | |
| 604 | + * offset past every <img> so a poster never ties ahead of one. | |
| 605 | + * | |
| 606 | + * @param string $html Full page HTML. | |
| 607 | + * @param string[] $exclusions Substring patterns the user excluded. | |
| 608 | + * @param array<int,array{0:int,1:int}> $skip_ranges Byte ranges of chrome containers. | |
| 609 | + * @return array<int,array{tag:string,src:string,srcset:string,sizes:string,score:float,order:int,background:bool}> | |
| 610 | + */ | |
| 611 | + private static function video_poster_candidates( string $html, array $exclusions, array $skip_ranges ): array { | |
| 612 | + if ( ! preg_match_all( '#<video\b[^>]*\bposter\s*=\s*(["\'])(.*?)\1[^>]*>#i', $html, $matches, PREG_OFFSET_CAPTURE ) ) { | |
| 613 | + return array(); | |
| 614 | + } | |
| 615 | + | |
| 616 | + $found = array(); | |
| 617 | + foreach ( $matches[0] as $index => $match ) { | |
| 618 | + [ $tag, $offset ] = $match; | |
| 619 | + $src = trim( html_entity_decode( $matches[2][ $index ][0], ENT_QUOTES ) ); | |
| 620 | + if ( '' === $src || 0 === stripos( $src, 'data:' ) ) { | |
| 621 | + continue; | |
| 622 | + } | |
| 623 | + if ( self::offset_in_ranges( $offset, $skip_ranges ) ) { | |
| 624 | + continue; | |
| 625 | + } | |
| 626 | + foreach ( $exclusions as $needle ) { | |
| 627 | + if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 628 | + continue 2; | |
| 629 | + } | |
| 630 | + } | |
| 631 | + if ( self::looks_too_small( $tag ) ) { | |
| 632 | + continue; | |
| 633 | + } | |
| 634 | + | |
| 635 | + $area = self::style_area( self::attr( $tag, 'style' ) ); | |
| 636 | + if ( 0 === $area ) { | |
| 637 | + $area = self::UNKNOWN_SIZE_SCORE; | |
| 638 | + } | |
| 639 | + | |
| 640 | + $found[] = array( | |
| 641 | + 'tag' => $tag, | |
| 642 | + 'src' => $src, | |
| 643 | + 'srcset' => '', | |
| 644 | + 'sizes' => '', | |
| 645 | + 'score' => (float) $area, | |
| 646 | + 'order' => 100000 + $index, | |
| 647 | + 'offset' => $offset, | |
| 648 | + 'background' => true, | |
| 649 | + ); | |
| 650 | + } | |
| 651 | + | |
| 652 | + return $found; | |
| 653 | + } | |
| 654 | + | |
| 655 | + /** | |
| 656 | + * How many <style>-block background rules are considered per page. The | |
| 657 | + * scan is linear, but each rule costs one class-lookup pass over the | |
| 658 | + * body, so a pathological page (thousands of generated rules) is capped | |
| 659 | + * rather than trusted. | |
| 660 | + */ | |
| 661 | + private const STYLE_RULE_BUDGET = 40; | |
| 662 | + | |
| 663 | + /** | |
| 664 | + * Collect background-image rules from inline <style> blocks whose | |
| 665 | + * selector matches an element in the body. | |
| 666 | + * | |
| 667 | + * The match is deliberately narrow: the rule's RIGHTMOST simple selector | |
| 668 | + * must carry a class or id, and the first element in the body bearing it | |
| 669 | + * (outside chrome containers) is taken as the painted element. Rules | |
| 670 | + * inside @media (or any other at-rule block) are skipped — a desktop-only | |
| 671 | + * background preloaded on mobile is a wasted high-priority fetch, and the | |
| 672 | + * markup gives no viewport to resolve the query against. | |
| 673 | + * | |
| 674 | + * @param string $html Full page HTML. | |
| 675 | + * @param string[] $exclusions Substring patterns the user excluded. | |
| 676 | + * @param array<int,array{0:int,1:int}> $skip_ranges Byte ranges of chrome containers. | |
| 677 | + * @return array<int,array{tag:string,src:string,srcset:string,sizes:string,score:float,order:int,background:bool}> | |
| 678 | + */ | |
| 679 | + private static function style_block_candidates( string $html, array $exclusions, array $skip_ranges ): array { | |
| 680 | + if ( ! preg_match_all( '#<style\b[^>]*>(.*?)</style\s*>#is', $html, $blocks ) ) { | |
| 681 | + return array(); | |
| 682 | + } | |
| 683 | + | |
| 684 | + $found = array(); | |
| 685 | + $budget = self::STYLE_RULE_BUDGET; | |
| 686 | + foreach ( $blocks[1] as $css ) { | |
| 687 | + if ( $budget <= 0 ) { | |
| 688 | + break; | |
| 689 | + } | |
| 690 | + $css = self::strip_at_rule_blocks( $css ); | |
| 691 | + if ( false === stripos( $css, 'url(' ) ) { | |
| 692 | + continue; | |
| 693 | + } | |
| 694 | + // One flat rule at a time: selector list up to '{', body to '}'. | |
| 695 | + if ( ! preg_match_all( '#(?:^|})\s*([^{}]{1,512})\{([^{}]*)\}#s', $css, $rules, PREG_SET_ORDER ) ) { | |
| 696 | + continue; | |
| 697 | + } | |
| 698 | + foreach ( $rules as $rule ) { | |
| 699 | + if ( $budget <= 0 ) { | |
| 700 | + break 2; | |
| 701 | + } | |
| 702 | + if ( false === stripos( $rule[2], 'url(' ) ) { | |
| 703 | + continue; | |
| 704 | + } | |
| 705 | + $src = self::background_url( $rule[2] ); | |
| 706 | + if ( '' === $src ) { | |
| 707 | + continue; | |
| 708 | + } | |
| 709 | + --$budget; | |
| 710 | + // First selector of the list, rightmost compound of it. | |
| 711 | + $selector = trim( (string) strtok( $rule[1], ',' ) ); | |
| 712 | + $parts = preg_split( '#[\s>+~]+#', $selector ); | |
| 713 | + $last = (string) end( $parts ); | |
| 714 | + // The last class or id token of that compound. Pseudo-classes | |
| 715 | + // (:hover, ::before) mean the background is not the initial | |
| 716 | + // paint, so they disqualify the rule. | |
| 717 | + if ( false !== strpos( $last, ':' ) ) { | |
| 718 | + continue; | |
| 719 | + } | |
| 720 | + if ( ! preg_match( '#([.\#])([-\w]+)$#', $last, $tok ) ) { | |
| 721 | + continue; | |
| 722 | + } | |
| 723 | + $el = '.' === $tok[1] | |
| 724 | + ? self::first_element_with_class( $html, $tok[2], $skip_ranges ) | |
| 725 | + : self::first_element_with_id( $html, $tok[2], $skip_ranges ); | |
| 726 | + if ( null === $el ) { | |
| 727 | + continue; | |
| 728 | + } | |
| 729 | + [ $tag, $offset ] = $el; | |
| 730 | + foreach ( $exclusions as $needle ) { | |
| 731 | + if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 732 | + continue 2; | |
| 733 | + } | |
| 734 | + } | |
| 735 | + if ( self::looks_too_small( $tag ) ) { | |
| 736 | + continue; | |
| 737 | + } | |
| 738 | + $area = self::style_area( self::attr( $tag, 'style' ) ); | |
| 739 | + if ( 0 === $area ) { | |
| 740 | + $area = self::UNKNOWN_SIZE_SCORE; | |
| 741 | + } | |
| 742 | + $found[] = array( | |
| 743 | + 'tag' => $tag, | |
| 744 | + 'src' => $src, | |
| 745 | + 'srcset' => '', | |
| 746 | + 'sizes' => '', | |
| 747 | + 'score' => (float) $area, | |
| 748 | + // Offset past the inline-style backgrounds: a rule-matched | |
| 749 | + // background is one inference step less certain, so it must | |
| 750 | + // never tie ahead of one read straight off the element. | |
| 751 | + 'order' => 200000 + $offset, | |
| 752 | + 'offset' => $offset, | |
| 753 | + 'background' => true, | |
| 754 | + ); | |
| 755 | + } | |
| 756 | + } | |
| 757 | + | |
| 758 | + return $found; | |
| 759 | + } | |
| 760 | + | |
| 761 | + /** | |
| 762 | + * CSS with every at-rule BLOCK (@media, @supports, @container, …) removed, | |
| 763 | + * by brace depth — a regex cannot pair nested braces. Flat at-rules | |
| 764 | + * (@import, @charset) have no block and pass through harmlessly. | |
| 765 | + */ | |
| 766 | + private static function strip_at_rule_blocks( string $css ): string { | |
| 767 | + $out = ''; | |
| 768 | + $len = strlen( $css ); | |
| 769 | + $i = 0; | |
| 770 | + while ( $i < $len ) { | |
| 771 | + $at = strpos( $css, '@', $i ); | |
| 772 | + if ( false === $at ) { | |
| 773 | + return $out . substr( $css, $i ); | |
| 774 | + } | |
| 775 | + $brace = strpos( $css, '{', $at ); | |
| 776 | + $semi = strpos( $css, ';', $at ); | |
| 777 | + $out .= substr( $css, $i, $at - $i ); | |
| 778 | + if ( false === $brace || ( false !== $semi && $semi < $brace ) ) { | |
| 779 | + // Flat at-rule — skip to its semicolon (or end). | |
| 780 | + $i = false === $semi ? $len : $semi + 1; | |
| 781 | + continue; | |
| 782 | + } | |
| 783 | + // Block at-rule — skip to its matching close brace. | |
| 784 | + $depth = 1; | |
| 785 | + $i = $brace + 1; | |
| 786 | + while ( $i < $len && $depth > 0 ) { | |
| 787 | + $c = $css[ $i ]; | |
| 788 | + if ( '{' === $c ) { | |
| 789 | + ++$depth; | |
| 790 | + } elseif ( '}' === $c ) { | |
| 791 | + --$depth; | |
| 792 | + } | |
| 793 | + ++$i; | |
| 794 | + } | |
| 795 | + } | |
| 796 | + return $out; | |
| 797 | + } | |
| 798 | + | |
| 799 | + /** | |
| 800 | + * The first element in the BODY carrying $class (outside chrome ranges), | |
| 801 | + * as [tag, offset], or null. Body-only, so a head <meta> can never match | |
| 802 | + * and a hit's offset is comparable with the chrome ranges. | |
| 803 | + * | |
| 804 | + * @return array{0:string,1:int}|null | |
| 805 | + */ | |
| 806 | + private static function first_element_with_class( string $html, string $class, array $skip_ranges ): ?array { | |
| 807 | + $body = stripos( $html, '<body' ); | |
| 808 | + $from = false === $body ? 0 : $body; | |
| 809 | + if ( ! preg_match_all( '#<[a-z][^>]*\bclass\s*=\s*(["\'])[^"\']*(?<![-\w])' . preg_quote( $class, '#' ) . '(?![-\w])[^"\']*\1[^>]*>#i', $html, $m, PREG_OFFSET_CAPTURE, $from ) ) { | |
| 810 | + return null; | |
| 811 | + } | |
| 812 | + foreach ( $m[0] as $match ) { | |
| 813 | + if ( ! self::offset_in_ranges( $match[1], $skip_ranges ) ) { | |
| 814 | + return array( $match[0], $match[1] ); | |
| 815 | + } | |
| 816 | + } | |
| 817 | + return null; | |
| 818 | + } | |
| 819 | + | |
| 820 | + /** | |
| 821 | + * The first element carrying id="$id" (outside chrome ranges), as | |
| 822 | + * [tag, offset], or null. | |
| 823 | + * | |
| 824 | + * @return array{0:string,1:int}|null | |
| 825 | + */ | |
| 826 | + private static function first_element_with_id( string $html, string $id, array $skip_ranges ): ?array { | |
| 827 | + $body = stripos( $html, '<body' ); | |
| 828 | + $from = false === $body ? 0 : $body; | |
| 829 | + if ( ! preg_match( '#<[a-z][^>]*\bid\s*=\s*(["\'])' . preg_quote( $id, '#' ) . '\1[^>]*>#i', $html, $m, PREG_OFFSET_CAPTURE, $from ) ) { | |
| 830 | + return null; | |
| 831 | + } | |
| 832 | + if ( self::offset_in_ranges( $m[0][1], $skip_ranges ) ) { | |
| 833 | + return null; | |
| 834 | + } | |
| 835 | + return array( $m[0][0], $m[0][1] ); | |
| 836 | + } | |
| 837 | + | |
| 838 | + /** | |
| 481 | 839 | * Pull a real image URL out of a `background`/`background-image` declaration. |
| 482 | 840 | * |
| 483 | 841 | * Returns '' for anything with nothing to fetch: a gradient (which is a |
| 484 | 842 | * background-image but not a resource), a data: URI, or `none`. |
| @@ -669,12 +1027,29 @@ | ||
| 669 | 1027 | } |
| 670 | 1028 | return $ranges; |
| 671 | 1029 | } |
| 672 | 1030 | |
| 1031 | + /** | |
| 1032 | + * Does a tag contain any of the given substring patterns? | |
| 1033 | + * | |
| 1034 | + * @param string[] $patterns | |
| 1035 | + */ | |
| 1036 | + private static function matches_any( string $tag, array $patterns ): bool { | |
| 1037 | + foreach ( $patterns as $needle ) { | |
| 1038 | + if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 1039 | + return true; | |
| 1040 | + } | |
| 1041 | + } | |
| 1042 | + return false; | |
| 1043 | + } | |
| 1044 | + | |
| 673 | 1045 | /** Does a byte offset fall inside any of the given [start, end] ranges? */ |
| 674 | 1046 | private static function offset_in_ranges( int $offset, array $ranges ): bool { |
| 675 | 1047 | foreach ( $ranges as $range ) { |
| 676 | - if ( $offset > $range[0] && $offset < $range[1] ) { | |
| 1048 | + // >= on the start: a closed <details> span from | |
| 1049 | + // Lazy_Loader::hidden_ranges() opens right at the first image | |
| 1050 | + // after its <summary>. | |
| 1051 | + if ( $offset >= $range[0] && $offset < $range[1] ) { | |
| 677 | 1052 | return true; |
| 678 | 1053 | } |
| 679 | 1054 | } |
| 680 | 1055 | return false; |
