| @@ -3,15 +3,18 @@ | ||
| 3 | 3 | * Resource Hints processor — pure HTML transformer for resource hints. |
| 4 | 4 | * |
| 5 | 5 | * Given a fully-rendered page and the Preload module's options, it: |
| 6 | 6 | * 1. Ranks every eligible <img> by the largest declared size it can read — |
| 7 | - * width×height attributes, else the widest srcset candidate — and emits | |
| 7 | + * width×height attributes, else the widest srcset candidate — boosted by | |
| 8 | + * the author's own priority signals (fetchpriority="high", an explicit | |
| 9 | + * loading="eager") and lightly weighted by document position, and emits | |
| 8 | 10 | * a <link rel="preload" as="image" fetchpriority="high"> for the top N |
| 9 | 11 | * in the <head>, carrying srcset/sizes as imagesrcset/imagesizes so the |
| 10 | 12 | * browser can pick the right candidate — then adds fetchpriority="high" |
| 11 | - * to the <img> itself so it beats any loading="lazy" the theme set. | |
| 12 | - * Ranking by size rather than document position: the first images on a | |
| 13 | - * real page are usually header chrome, not the hero (#96). | |
| 13 | + * to the <img> itself. Ranking by size rather than document position: | |
| 14 | + * the first images on a real page are usually header chrome, not the | |
| 15 | + * hero (#96). Images inside <footer>/<nav>/<aside>, images the theme | |
| 16 | + * explicitly lazy-loads, and tiny images never compete (FBS-84576). | |
| 14 | 17 | * 2. Emits <link rel="preconnect"> for detected web-font hosts |
| 15 | 18 | * (fonts.googleapis.com + fonts.gstatic.com) and any user-supplied |
| 16 | 19 | * hosts, deduped. |
| 17 | 20 | * |
| @@ -60,8 +63,20 @@ | ||
| 60 | 63 | [ $html, $preload ] = self::build_lcp_preload( $html, $count, $exclusions ); |
| 61 | 64 | $hints .= $preload; |
| 62 | 65 | } |
| 63 | 66 | |
| 67 | + // The manual list runs AFTER the automatic pick so it can deduplicate | |
| 68 | + // against it: a URL both name should carry ONE hint, not two. It exists | |
| 69 | + // for the image the detector cannot see — most often a hero section's | |
| 70 | + // CSS background-image, where the LCP element is a <div> with none of | |
| 71 | + // the width/height/fetchpriority signals the scorer reads. On the site | |
| 72 | + // that surfaced this, 3.3s of a 5.3s mobile LCP was pure discovery | |
| 73 | + // delay for exactly such an image. (FBS-84578) | |
| 74 | + $manual = array_filter( array_map( 'strval', (array) ( $opts['preload_images'] ?? array() ) ) ); | |
| 75 | + if ( ! empty( $manual ) ) { | |
| 76 | + $hints .= self::build_manual_image_preloads( $manual, $hints ); | |
| 77 | + } | |
| 78 | + | |
| 64 | 79 | // Full-page eager promotion for Lazy-excluded heroes (FBS-83553 H2). The |
| 65 | 80 | // Lazy module only filters the_content/thumbnail/avatar/widget, so a |
| 66 | 81 | // theme/builder hero printed OUTSIDE those keeps WP core's |
| 67 | 82 | // loading="lazy". This pass runs over the whole document, so it can reach |
| @@ -80,26 +95,90 @@ | ||
| 80 | 95 | return self::inject_into_head( $html, $hints ); |
| 81 | 96 | } |
| 82 | 97 | |
| 83 | 98 | /** |
| 84 | - * Strip core `loading="lazy"` and add `fetchpriority="high"` + | |
| 85 | - * `decoding="async"` on every <img> whose tag matches one of the given | |
| 86 | - * exclusion substrings. Mirrors what Lazy_Loader does for an excluded image | |
| 87 | - * inside the_content, but page-wide so heroes outside it are covered too. | |
| 88 | - * (FBS-83553 H2) | |
| 99 | + * How many entries of the manual preload list are honoured. Preloading | |
| 100 | + * competes with the page for its top network priority — a long list | |
| 101 | + * inverts the benefit, so the cap is deliberately small. | |
| 102 | + */ | |
| 103 | + private const MAX_MANUAL_PRELOADS = 3; | |
| 104 | + | |
| 105 | + /** | |
| 106 | + * One `<link rel="preload" as="image">` per manual list entry. | |
| 89 | 107 | * |
| 108 | + * Entries are full URLs or site-relative paths. Anything that is neither | |
| 109 | + * (a data: URI, a bare word) is skipped rather than guessed at, and a URL | |
| 110 | + * the automatic pick already emitted is skipped too — one hint per image, | |
| 111 | + * whoever names it first. | |
| 112 | + * | |
| 113 | + * @param array<int,string> $urls The configured list. | |
| 114 | + * @param string $existing_hints Hints already built this request. | |
| 115 | + */ | |
| 116 | + private static function build_manual_image_preloads( array $urls, string $existing_hints ): string { | |
| 117 | + /** | |
| 118 | + * Filter the manual image-preload list before it is emitted. | |
| 119 | + * | |
| 120 | + * @param array<int,string> $urls Configured URLs, in panel order. | |
| 121 | + */ | |
| 122 | + $urls = (array) apply_filters( 'xspeed_preload_images', $urls ); | |
| 123 | + | |
| 124 | + $out = ''; | |
| 125 | + $seen = array(); | |
| 126 | + foreach ( $urls as $url ) { | |
| 127 | + $url = trim( (string) $url ); | |
| 128 | + if ( '' === $url ) { | |
| 129 | + continue; | |
| 130 | + } | |
| 131 | + // A full URL or a site-relative path; nothing else is guessable. | |
| 132 | + $is_absolute = (bool) preg_match( '#^https?://#i', $url ); | |
| 133 | + $is_relative = '' !== $url && '/' === $url[0] && ( strlen( $url ) < 2 || '/' !== $url[1] ); | |
| 134 | + if ( ! $is_absolute && ! $is_relative ) { | |
| 135 | + continue; | |
| 136 | + } | |
| 137 | + $href = esc_url( $url ); | |
| 138 | + if ( '' === $href || isset( $seen[ $href ] ) || false !== strpos( $existing_hints, 'href="' . $href . '"' ) ) { | |
| 139 | + continue; | |
| 140 | + } | |
| 141 | + $seen[ $href ] = true; | |
| 142 | + $out .= '<link rel="preload" as="image" href="' . $href . '" fetchpriority="high">'; | |
| 143 | + if ( count( $seen ) >= self::MAX_MANUAL_PRELOADS ) { | |
| 144 | + break; | |
| 145 | + } | |
| 146 | + } | |
| 147 | + | |
| 148 | + return $out; | |
| 149 | + } | |
| 150 | + | |
| 151 | + /** | |
| 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) | |
| 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 | + * | |
| 90 | 162 | * @param string $html Full page HTML. |
| 91 | 163 | * @param string[] $exclusions Substring patterns identifying above-the-fold heroes. |
| 92 | 164 | */ |
| 93 | 165 | private static function promote_excluded_images( string $html, array $exclusions ): string { |
| 94 | - return (string) preg_replace_callback( | |
| 95 | - '#<img\b(?:"[^"]*"|\'[^\']*\'|[^>"\'])*>#i', | |
| 96 | - static function ( array $m ) use ( $exclusions ) { | |
| 97 | - $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]; | |
| 98 | 174 | foreach ( $exclusions as $needle ) { |
| 99 | 175 | if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { |
| 100 | 176 | $tag = (string) preg_replace( '#\s*\bloading=(["\'])\s*lazy\s*\1#i', '', $tag ); |
| 101 | - $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 | + } | |
| 102 | 181 | if ( ! preg_match( '#\bdecoding=#i', $tag ) ) { |
| 103 | 182 | $tag = (string) preg_replace( '#<img\b#i', '<img decoding="async"', $tag, 1 ); |
| 104 | 183 | } |
| 105 | 184 | return $tag; |
| @@ -106,10 +185,15 @@ | ||
| 106 | 185 | } |
| 107 | 186 | } |
| 108 | 187 | return $tag; |
| 109 | 188 | }, |
| 110 | - $html | |
| 189 | + $html, | |
| 190 | + -1, | |
| 191 | + $count, | |
| 192 | + PREG_OFFSET_CAPTURE | |
| 111 | 193 | ); |
| 194 | + | |
| 195 | + return is_string( $result ) ? $result : $html; | |
| 112 | 196 | } |
| 113 | 197 | |
| 114 | 198 | /** |
| 115 | 199 | * Build preconnect <link>s for detected font hosts + user hosts. |
| @@ -187,23 +271,43 @@ | ||
| 187 | 271 | // nothing — it just adds a high-priority request competing with the |
| 188 | 272 | // one that matters, and the feature reported success either way. The |
| 189 | 273 | // marker list and size gate were heuristics layered on top of the |
| 190 | 274 | // wrong primitive rather than replacing it. (#96) |
| 191 | - $candidates = array(); | |
| 192 | - if ( preg_match_all( '#<img\b[^>]*>#i', $html, $matches ) ) { | |
| 193 | - foreach ( $matches[0] as $index => $tag ) { | |
| 275 | + $candidates = array(); | |
| 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(); | |
| 283 | + if ( preg_match_all( '#<img\b[^>]*>#i', $html, $matches, PREG_OFFSET_CAPTURE ) ) { | |
| 284 | + foreach ( $matches[0] as $index => $match ) { | |
| 285 | + [ $tag, $offset ] = $match; | |
| 286 | + | |
| 194 | 287 | // Skip anything the user excluded. |
| 195 | - $excluded = false; | |
| 196 | - foreach ( $exclusions as $needle ) { | |
| 197 | - if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 198 | - $excluded = true; | |
| 199 | - break; | |
| 200 | - } | |
| 288 | + if ( self::matches_any( $tag, $exclusions ) ) { | |
| 289 | + continue; | |
| 201 | 290 | } |
| 202 | - if ( $excluded ) { | |
| 291 | + | |
| 292 | + // An image inside <footer>/<nav>/<aside> is site chrome by | |
| 293 | + // construction — a footer brand strip or FAQ illustration can | |
| 294 | + // never be the LCP element, whatever size it declares. On the | |
| 295 | + // FBS-84576 repro these decoys outranked the real hero three | |
| 296 | + // times on one layout. | |
| 297 | + if ( self::offset_in_ranges( $offset, $skip_ranges ) || Lazy_Loader::tag_is_hidden( $tag, 'img' ) ) { | |
| 298 | + $ruled_out[ $index ] = true; | |
| 203 | 299 | continue; |
| 204 | 300 | } |
| 205 | 301 | |
| 302 | + // An image the theme explicitly lazy-loads is never the | |
| 303 | + // intended LCP — the author has already said "this can wait". | |
| 304 | + // Preloading it would contradict the markup and steal | |
| 305 | + // bandwidth from the image that matters. (FBS-84576) | |
| 306 | + if ( 'lazy' === strtolower( self::attr( $tag, 'loading' ) ) ) { | |
| 307 | + continue; | |
| 308 | + } | |
| 309 | + | |
| 206 | 310 | // Resolve the EFFECTIVE image URL. Page builders + JS lazy |
| 207 | 311 | // loaders park a placeholder (a data: URI or a 1px spacer) in |
| 208 | 312 | // `src` and the real URL in `data-src`, so the hero the browser |
| 209 | 313 | // actually paints is behind data-src. (FBS-83553 H1) |
| @@ -214,8 +318,9 @@ | ||
| 214 | 318 | |
| 215 | 319 | // Chrome markers / explicit opt-out / obviously-tiny images |
| 216 | 320 | // never compete. (FBS-83553 H1 "logo before hero".) |
| 217 | 321 | if ( self::looks_too_small( $tag ) ) { |
| 322 | + $ruled_out[ $index ] = true; | |
| 218 | 323 | continue; |
| 219 | 324 | } |
| 220 | 325 | |
| 221 | 326 | $candidates[] = array( |
| @@ -222,15 +327,66 @@ | ||
| 222 | 327 | 'tag' => $tag, |
| 223 | 328 | 'src' => $src, |
| 224 | 329 | 'srcset' => $srcset, |
| 225 | 330 | 'sizes' => $sizes, |
| 226 | - 'score' => self::lcp_score( $tag, $srcset ), | |
| 331 | + 'score' => self::weighted_score( self::lcp_score( $tag, $srcset ), $tag, $index ), | |
| 227 | 332 | 'order' => $index, |
| 333 | + 'offset' => $offset, | |
| 228 | 334 | ); |
| 229 | 335 | } |
| 230 | 336 | } |
| 231 | 337 | |
| 338 | + // PASS 1b — the same for CSS background images. | |
| 339 | + // | |
| 340 | + // On a page builder the hero is usually a background-image on the | |
| 341 | + // section, not an <img>, so an <img>-only candidate set never contained | |
| 342 | + // the element that actually paints as LCP. It preloaded whatever <img> | |
| 343 | + // happened to be there — measured at 0ms against the feature switched | |
| 344 | + // off, while spending a high-priority fetch on the critical path — or, | |
| 345 | + // on a page with no <img> at all, emitted nothing. (#247) | |
| 346 | + foreach ( self::background_candidates( $html, $exclusions, $skip_ranges ) as $bg ) { | |
| 347 | + $candidates[] = $bg; | |
| 348 | + } | |
| 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 | + | |
| 232 | 385 | if ( empty( $candidates ) ) { |
| 386 | + if ( null !== $video_at ) { | |
| 387 | + $html = self::demote_images( $html, $ruled_out ); | |
| 388 | + } | |
| 233 | 389 | return array( $html, '' ); |
| 234 | 390 | } |
| 235 | 391 | |
| 236 | 392 | // Rank by score, biggest first. Document order breaks ties, so two |
| @@ -261,9 +417,15 @@ | ||
| 261 | 417 | ); |
| 262 | 418 | if ( ! $already ) { |
| 263 | 419 | $preload .= self::preload_link( $w['src'], $w['srcset'], $w['sizes'] ); |
| 264 | 420 | } |
| 265 | - $chosen[ $w['order'] ] = true; | |
| 421 | + // Only <img> winners are promoted in PASS 2 — there is no | |
| 422 | + // fetchpriority/loading attribute to fix on a background element, | |
| 423 | + // and its `order` is offset past every <img> index precisely so it | |
| 424 | + // can never select one for rewriting. | |
| 425 | + if ( empty( $w['background'] ) ) { | |
| 426 | + $chosen[ $w['order'] ] = true; | |
| 427 | + } | |
| 266 | 428 | } |
| 267 | 429 | |
| 268 | 430 | // Rewrite only the winning tags. Counting occurrences rather than |
| 269 | 431 | // matching on tag text, because the same markup can legitimately |
| @@ -268,14 +430,26 @@ | ||
| 268 | 430 | // Rewrite only the winning tags. Counting occurrences rather than |
| 269 | 431 | // matching on tag text, because the same markup can legitimately |
| 270 | 432 | // appear more than once on a page and only the ranked instance should |
| 271 | 433 | // be promoted. |
| 272 | - $seen = -1; | |
| 273 | - $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( | |
| 274 | 445 | '#<img\b[^>]*>#i', |
| 275 | - static function ( array $m ) use ( &$seen, $chosen ) { | |
| 446 | + static function ( array $m ) use ( &$seen, $chosen, $demote ) { | |
| 276 | 447 | ++$seen; |
| 277 | 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 | + } | |
| 278 | 452 | return $m[0]; |
| 279 | 453 | } |
| 280 | 454 | // Add fetchpriority="high" AND remove any loading="lazy" the |
| 281 | 455 | // theme / WP core left on the LCP image. fetchpriority="high" |
| @@ -290,8 +464,432 @@ | ||
| 290 | 464 | return array( (string) $html, $preload ); |
| 291 | 465 | } |
| 292 | 466 | |
| 293 | 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 | + /** | |
| 516 | + * Collect CSS `background-image` heroes as LCP candidates. | |
| 517 | + * | |
| 518 | + * Only INLINE `style` attributes are read. A background declared in an | |
| 519 | + * external stylesheet is invisible here by design: resolving it would mean | |
| 520 | + * fetching and parsing CSS from inside an output-buffer pass, and the URL a | |
| 521 | + * selector resolves to depends on cascade order we cannot evaluate from | |
| 522 | + * markup. Builders that put the hero in a generated per-post stylesheet are | |
| 523 | + * therefore still unserved — worth doing, but not at this cost. (#247) | |
| 524 | + * | |
| 525 | + * Scores are the element's declared pixel area so a background competes | |
| 526 | + * against an <img> in the SAME units — the whole point being that the | |
| 527 | + * bigger of the two should win regardless of which kind it is. | |
| 528 | + * | |
| 529 | + * @param string $html Full page HTML. | |
| 530 | + * @param string[] $exclusions Substring patterns the user excluded. | |
| 531 | + * @param array<int,array{0:int,1:int}> $skip_ranges Byte ranges of chrome containers. | |
| 532 | + * @return array<int,array{tag:string,src:string,srcset:string,sizes:string,score:float,order:int,background:bool}> | |
| 533 | + */ | |
| 534 | + private static function background_candidates( string $html, array $exclusions, array $skip_ranges ): array { | |
| 535 | + if ( ! preg_match_all( '#<(?:div|section|header|figure|a|span|li|main|article|aside)\b[^>]*\sstyle\s*=\s*(["\']).*?\1[^>]*>#is', $html, $matches, PREG_OFFSET_CAPTURE ) ) { | |
| 536 | + return array(); | |
| 537 | + } | |
| 538 | + | |
| 539 | + $found = array(); | |
| 540 | + foreach ( $matches[0] as $index => $match ) { | |
| 541 | + [ $tag, $offset ] = $match; | |
| 542 | + | |
| 543 | + // The same chrome-container gate as <img>: a background painted | |
| 544 | + // inside <footer>/<nav>/<aside> is never the hero. (FBS-84576) | |
| 545 | + if ( self::offset_in_ranges( $offset, $skip_ranges ) ) { | |
| 546 | + continue; | |
| 547 | + } | |
| 548 | + | |
| 549 | + $style = self::attr( $tag, 'style' ); | |
| 550 | + if ( '' === $style || false === stripos( $style, 'background' ) ) { | |
| 551 | + continue; | |
| 552 | + } | |
| 553 | + | |
| 554 | + $src = self::background_url( $style ); | |
| 555 | + if ( '' === $src ) { | |
| 556 | + continue; | |
| 557 | + } | |
| 558 | + | |
| 559 | + foreach ( $exclusions as $needle ) { | |
| 560 | + if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 561 | + continue 2; | |
| 562 | + } | |
| 563 | + } | |
| 564 | + | |
| 565 | + // Same chrome/opt-out gates as <img>. A logo painted as a background | |
| 566 | + // is no more the hero than a logo in an <img>. | |
| 567 | + if ( self::looks_too_small( $tag ) ) { | |
| 568 | + continue; | |
| 569 | + } | |
| 570 | + | |
| 571 | + $area = self::style_area( $style ); | |
| 572 | + if ( 0 === $area ) { | |
| 573 | + // Nothing readable. Deliberately non-zero for the same reason | |
| 574 | + // UNKNOWN_SIZE_SCORE is: an unmeasurable background must still | |
| 575 | + // beat nothing on a page that declares no sizes at all, while | |
| 576 | + // losing to anything we can actually measure. | |
| 577 | + $area = self::UNKNOWN_SIZE_SCORE; | |
| 578 | + } | |
| 579 | + | |
| 580 | + $found[] = array( | |
| 581 | + 'tag' => $tag, | |
| 582 | + 'src' => $src, | |
| 583 | + 'srcset' => '', | |
| 584 | + 'sizes' => '', | |
| 585 | + 'score' => (float) $area, | |
| 586 | + // Offset so a background never ties ahead of an <img> that | |
| 587 | + // appeared earlier in the document; ties still break on order. | |
| 588 | + 'order' => 100000 + $index, | |
| 589 | + 'offset' => $offset, | |
| 590 | + 'background' => true, | |
| 591 | + ); | |
| 592 | + } | |
| 593 | + | |
| 594 | + return $found; | |
| 595 | + } | |
| 596 | + | |
| 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 | + /** | |
| 839 | + * Pull a real image URL out of a `background`/`background-image` declaration. | |
| 840 | + * | |
| 841 | + * Returns '' for anything with nothing to fetch: a gradient (which is a | |
| 842 | + * background-image but not a resource), a data: URI, or `none`. | |
| 843 | + */ | |
| 844 | + private static function background_url( string $style ): string { | |
| 845 | + // Decode BEFORE parsing. Builders emit the url() quotes HTML-encoded | |
| 846 | + // inside a style attribute (url("/hero.jpg")), and `"` | |
| 847 | + // carries a semicolon — so splitting the declaration on `;` first | |
| 848 | + // truncated the value to `url("` and found no URL at all. | |
| 849 | + $style = html_entity_decode( $style, ENT_QUOTES ); | |
| 850 | + | |
| 851 | + if ( ! preg_match( '#background(?:-image)?\s*:\s*((?:[^;\'"]|"[^"]*"|\'[^\']*\')+)#i', $style, $decl ) ) { | |
| 852 | + return ''; | |
| 853 | + } | |
| 854 | + if ( ! preg_match( '#url\(\s*(["\']?)(.*?)\1\s*\)#is', $decl[1], $m ) ) { | |
| 855 | + return ''; | |
| 856 | + } | |
| 857 | + $url = trim( $m[2] ); | |
| 858 | + if ( '' === $url || 0 === stripos( $url, 'data:' ) ) { | |
| 859 | + return ''; | |
| 860 | + } | |
| 861 | + return $url; | |
| 862 | + } | |
| 863 | + | |
| 864 | + /** | |
| 865 | + * Declared pixel area from an inline style, or 0 when it can't be read. | |
| 866 | + * | |
| 867 | + * Only px is honoured. A percentage or viewport unit resolves against a | |
| 868 | + * containing block we cannot see from markup, and guessing one produced the | |
| 869 | + * wrong winner more often than declining to. | |
| 870 | + */ | |
| 871 | + private static function style_area( string $style ): int { | |
| 872 | + $w = self::style_px( $style, 'width' ); | |
| 873 | + $h = self::style_px( $style, 'height' ); | |
| 874 | + if ( $w > 0 && $h > 0 ) { | |
| 875 | + return $w * $h; | |
| 876 | + } | |
| 877 | + if ( $w > 0 ) { | |
| 878 | + return (int) round( $w * $w * self::ASSUMED_ASPECT_RATIO ); | |
| 879 | + } | |
| 880 | + return 0; | |
| 881 | + } | |
| 882 | + | |
| 883 | + /** One px-valued CSS length from an inline style, or 0. */ | |
| 884 | + private static function style_px( string $style, string $prop ): int { | |
| 885 | + if ( preg_match( '#(?:^|;)\s*' . preg_quote( $prop, '#' ) . '\s*:\s*(\d+(?:\.\d+)?)px#i', $style, $m ) ) { | |
| 886 | + return (int) round( (float) $m[1] ); | |
| 887 | + } | |
| 888 | + return 0; | |
| 889 | + } | |
| 890 | + | |
| 891 | + /** | |
| 294 | 892 | * How likely is this <img> to be the LCP element? Higher wins. |
| 295 | 893 | * |
| 296 | 894 | * Rendered area is the best available proxy, and we can only read what the |
| 297 | 895 | * markup declares: |
| @@ -307,13 +905,13 @@ | ||
| 307 | 905 | * |
| 308 | 906 | * @param string $tag The full <img> tag. |
| 309 | 907 | * @param string $srcset Resolved srcset (may come from data-srcset). |
| 310 | 908 | */ |
| 311 | - private static function lcp_score( string $tag, string $srcset ): int { | |
| 909 | + private static function lcp_score( string $tag, string $srcset ): float { | |
| 312 | 910 | $w = self::attr( $tag, 'width' ); |
| 313 | 911 | $h = self::attr( $tag, 'height' ); |
| 314 | 912 | if ( '' !== $w && '' !== $h && is_numeric( $w ) && is_numeric( $h ) ) { |
| 315 | - return (int) $w * (int) $h; | |
| 913 | + return (float) ( (int) $w * (int) $h ); | |
| 316 | 914 | } |
| 317 | 915 | |
| 318 | 916 | $widest = self::widest_srcset_width( $srcset ); |
| 319 | 917 | if ( $widest > 0 ) { |
| @@ -323,15 +921,142 @@ | ||
| 323 | 921 | // 1024w sidebar thumbnail (1 048 576) beat a declared 1200×600 |
| 324 | 922 | // hero (720 000) — a regression on exactly the mixed pages that |
| 325 | 923 | // document order used to get right, since the hero usually comes |
| 326 | 924 | // first. Assuming a 16:9 box keeps both sides in the same units. |
| 327 | - return (int) round( $widest * $widest * self::ASSUMED_ASPECT_RATIO ); | |
| 925 | + return round( $widest * $widest * self::ASSUMED_ASPECT_RATIO ); | |
| 328 | 926 | } |
| 329 | 927 | |
| 330 | - return self::UNKNOWN_SIZE_SCORE; | |
| 928 | + return (float) self::UNKNOWN_SIZE_SCORE; | |
| 331 | 929 | } |
| 332 | 930 | |
| 333 | 931 | /** |
| 932 | + * Fold the author's own priority signals and document position into an | |
| 933 | + * area score. (FBS-84576) | |
| 934 | + * | |
| 935 | + * Boosts are ADDITIVE, in area units, so they can rescue an image whose | |
| 936 | + * size the markup doesn't declare: a hero with no width/height and no | |
| 937 | + * `w`-descriptor srcset scores UNKNOWN_SIZE_SCORE, and multiplying that | |
| 938 | + * by any factor still loses to a 548×136 logo that declares itself. This | |
| 939 | + * is exactly the live miss — the real hero carried loading="eager" | |
| 940 | + * fetchpriority="high" and lost to three dimension-declaring decoys. | |
| 941 | + * | |
| 942 | + * - fetchpriority="high" is the strongest signal there is: the author | |
| 943 | + * (or WP core's own LCP detection) has already named this image the | |
| 944 | + * hero. Worth a hero-sized area. | |
| 945 | + * - An EXPLICIT loading="eager" is a weaker but deliberate "load me | |
| 946 | + * now" (the default is eager, so writing it out is a choice). | |
| 947 | + * | |
| 948 | + * Position is a light multiplicative weight — earlier is better, but the | |
| 949 | + * spread is capped well under 5× so it can only break near-ties, never | |
| 950 | + * outrank a genuinely larger image further down (the logo-vs-hero case). | |
| 951 | + * | |
| 952 | + * @param float $score Base area score from lcp_score() / style_area(). | |
| 953 | + * @param string $tag The candidate's full tag (for the signal attrs). | |
| 954 | + * @param int $order Document-order index of the candidate. | |
| 955 | + */ | |
| 956 | + private static function weighted_score( float $score, string $tag, int $order ): float { | |
| 957 | + if ( 'high' === strtolower( self::attr( $tag, 'fetchpriority' ) ) ) { | |
| 958 | + $score += self::FETCHPRIORITY_HIGH_BOOST; | |
| 959 | + } | |
| 960 | + if ( 'eager' === strtolower( self::attr( $tag, 'loading' ) ) ) { | |
| 961 | + $score += self::EAGER_BOOST; | |
| 962 | + } | |
| 963 | + return $score * self::position_weight( $order ); | |
| 964 | + } | |
| 965 | + | |
| 966 | + /** | |
| 967 | + * Document-position weight: 1.25 for the first image, easing to 1.0 by | |
| 968 | + * the tenth. The whole spread is 25%, far under the 5× area difference it | |
| 969 | + * must never override — it exists only to keep the old first-wins | |
| 970 | + * behaviour for images we can't tell apart. | |
| 971 | + */ | |
| 972 | + private static function position_weight( int $order ): float { | |
| 973 | + return 1.0 + 0.25 * max( 0.0, 1.0 - $order / 10 ); | |
| 974 | + } | |
| 975 | + | |
| 976 | + /** | |
| 977 | + * Area-unit boost for fetchpriority="high" — roughly a 940×530 hero, so | |
| 978 | + * an explicitly-marked image outranks any mid-page decoy even when its | |
| 979 | + * own size is unreadable, while a genuinely huge unmarked image can still | |
| 980 | + * beat a marked small one. | |
| 981 | + */ | |
| 982 | + private const FETCHPRIORITY_HIGH_BOOST = 500000.0; | |
| 983 | + | |
| 984 | + /** | |
| 985 | + * Area-unit boost for an explicit loading="eager" — roughly 420×240, | |
| 986 | + * enough to break ties in favour of the author's intent without letting | |
| 987 | + * an eager logo outrank a plain hero. | |
| 988 | + */ | |
| 989 | + private const EAGER_BOOST = 100000.0; | |
| 990 | + | |
| 991 | + /** | |
| 992 | + * Byte ranges of <footer>/<nav>/<aside> regions. Nesting-aware per tag | |
| 993 | + * name (a nav inside a nav extends the range); an unclosed open tag | |
| 994 | + * poisons through to the end of the document, which errs on the side of | |
| 995 | + * not preloading — the safe direction, since a wrong preload is worse | |
| 996 | + * than none. (FBS-84576) | |
| 997 | + * | |
| 998 | + * @return array<int,array{0:int,1:int}> [start, end] byte offsets. | |
| 999 | + */ | |
| 1000 | + private static function chrome_container_ranges( string $html ): array { | |
| 1001 | + $ranges = array(); | |
| 1002 | + foreach ( array( 'footer', 'nav', 'aside' ) as $name ) { | |
| 1003 | + // (?=[\s/>]) rather than \b: a word boundary sits before the `-` | |
| 1004 | + // of a custom element, so `<nav\b` would swallow `<nav-menu>`. | |
| 1005 | + if ( ! preg_match_all( '#<(/?)' . $name . '(?=[\s/>])[^>]*>#i', $html, $m, PREG_OFFSET_CAPTURE ) ) { | |
| 1006 | + continue; | |
| 1007 | + } | |
| 1008 | + $depth = 0; | |
| 1009 | + $start = 0; | |
| 1010 | + foreach ( $m[0] as $i => $match ) { | |
| 1011 | + $closing = '' !== $m[1][ $i ][0]; | |
| 1012 | + if ( ! $closing ) { | |
| 1013 | + if ( 0 === $depth ) { | |
| 1014 | + $start = $match[1]; | |
| 1015 | + } | |
| 1016 | + ++$depth; | |
| 1017 | + } elseif ( $depth > 0 ) { | |
| 1018 | + --$depth; | |
| 1019 | + if ( 0 === $depth ) { | |
| 1020 | + $ranges[] = array( $start, $match[1] ); | |
| 1021 | + } | |
| 1022 | + } | |
| 1023 | + } | |
| 1024 | + if ( $depth > 0 ) { | |
| 1025 | + $ranges[] = array( $start, strlen( $html ) ); | |
| 1026 | + } | |
| 1027 | + } | |
| 1028 | + return $ranges; | |
| 1029 | + } | |
| 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 | + | |
| 1045 | + /** Does a byte offset fall inside any of the given [start, end] ranges? */ | |
| 1046 | + private static function offset_in_ranges( int $offset, array $ranges ): bool { | |
| 1047 | + foreach ( $ranges as $range ) { | |
| 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] ) { | |
| 1052 | + return true; | |
| 1053 | + } | |
| 1054 | + } | |
| 1055 | + return false; | |
| 1056 | + } | |
| 1057 | + | |
| 1058 | + /** | |
| 334 | 1059 | * Score for an image whose size we can't read at all. |
| 335 | 1060 | * |
| 336 | 1061 | * Deliberately non-zero: an unmeasurable image must still beat nothing and |
| 337 | 1062 | * still be preloadable on a page where no image declares its size. But it |
| @@ -492,13 +1217,21 @@ | ||
| 492 | 1217 | */ |
| 493 | 1218 | private const NON_HERO_MARKERS = array( 'logo', 'icon', 'avatar', 'gravatar', 'spinner', 'emoji', 'site-icon', 'custom-logo' ); |
| 494 | 1219 | |
| 495 | 1220 | /** |
| 1221 | + * Below this declared area (px²) an image is a badge/thumb/divider, never | |
| 1222 | + * an LCP hero — 10 000 is a 100×100 square, or a 500×20 strip. Applied | |
| 1223 | + * only when BOTH dimensions are readable. (FBS-84576) | |
| 1224 | + */ | |
| 1225 | + private const MIN_LCP_AREA = 10000; | |
| 1226 | + | |
| 1227 | + /** | |
| 496 | 1228 | * Is this <img> too small / too chrome-like to be the LCP hero? True when |
| 497 | 1229 | * either (a) it carries a logo/icon/avatar marker, (b) an explicit |
| 498 | - * `data-no-lcp` opt-out, or (c) BOTH width and height are present and both | |
| 499 | - * are ≤ the threshold. Missing dimensions are NOT guessed — an image whose | |
| 500 | - * size we can't read still competes. (FBS-83553 H1 "logo before hero".) | |
| 1230 | + * `data-no-lcp` opt-out, or (c) BOTH width and height are present and | |
| 1231 | + * both are ≤ the dimension threshold, or their area is under | |
| 1232 | + * MIN_LCP_AREA. Missing dimensions are NOT guessed — an image whose size | |
| 1233 | + * we can't read still competes. (FBS-83553 H1 "logo before hero".) | |
| 501 | 1234 | */ |
| 502 | 1235 | private static function looks_too_small( string $tag ): bool { |
| 503 | 1236 | if ( false !== stripos( $tag, 'data-no-lcp' ) ) { |
| 504 | 1237 | return true; |
| @@ -514,8 +1247,11 @@ | ||
| 514 | 1247 | $w = self::attr( $tag, 'width' ); |
| 515 | 1248 | $h = self::attr( $tag, 'height' ); |
| 516 | 1249 | if ( '' === $w || '' === $h || ! is_numeric( $w ) || ! is_numeric( $h ) ) { |
| 517 | 1250 | return false; // unknown size — don't guess; let it compete. |
| 1251 | + } | |
| 1252 | + if ( (int) $w * (int) $h < self::MIN_LCP_AREA ) { | |
| 1253 | + return true; | |
| 518 | 1254 | } |
| 519 | 1255 | return (int) $w <= self::MIN_LCP_DIMENSION && (int) $h <= self::MIN_LCP_DIMENSION; |
| 520 | 1256 | } |
| 521 | 1257 | |
