| @@ -2,13 +2,19 @@ | ||
| 2 | 2 | /** |
| 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 | - * 1. Finds the first N above-the-fold <img> (the LCP candidate) and emits | |
| 7 | - * a <link rel="preload" as="image" fetchpriority="high"> for each in | |
| 8 | - * the <head>, carrying srcset/sizes as imagesrcset/imagesizes so the | |
| 6 | + * 1. Ranks every eligible <img> by the largest declared size it can read — | |
| 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 | |
| 10 | + * a <link rel="preload" as="image" fetchpriority="high"> for the top N | |
| 11 | + * in the <head>, carrying srcset/sizes as imagesrcset/imagesizes so the | |
| 9 | 12 | * browser can pick the right candidate — then adds fetchpriority="high" |
| 10 | - * to the <img> itself so it beats any loading="lazy" the theme set. | |
| 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). | |
| 11 | 17 | * 2. Emits <link rel="preconnect"> for detected web-font hosts |
| 12 | 18 | * (fonts.googleapis.com + fonts.gstatic.com) and any user-supplied |
| 13 | 19 | * hosts, deduped. |
| 14 | 20 | * |
| @@ -53,12 +59,24 @@ | ||
| 53 | 59 | |
| 54 | 60 | if ( ! empty( $opts['lcp_preload'] ) ) { |
| 55 | 61 | $count = max( 0, (int) ( $opts['lcp_image_count'] ?? 1 ) ); |
| 56 | 62 | $exclusions = array_filter( array_map( 'strval', (array) ( $opts['lcp_exclusions'] ?? array() ) ) ); |
| 57 | - [ $html, $preload ] = self::build_lcp_preload( $html, $count, $exclusions ); | |
| 63 | + [ $html, $preload ] = self::build_lcp_preload( $html, $count, $exclusions, (string) ( $opts['page_url'] ?? '' ) ); | |
| 58 | 64 | $hints .= $preload; |
| 59 | 65 | } |
| 60 | 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 | + | |
| 61 | 79 | // Full-page eager promotion for Lazy-excluded heroes (FBS-83553 H2). The |
| 62 | 80 | // Lazy module only filters the_content/thumbnail/avatar/widget, so a |
| 63 | 81 | // theme/builder hero printed OUTSIDE those keeps WP core's |
| 64 | 82 | // loading="lazy". This pass runs over the whole document, so it can reach |
| @@ -77,26 +95,90 @@ | ||
| 77 | 95 | return self::inject_into_head( $html, $hints ); |
| 78 | 96 | } |
| 79 | 97 | |
| 80 | 98 | /** |
| 81 | - * Strip core `loading="lazy"` and add `fetchpriority="high"` + | |
| 82 | - * `decoding="async"` on every <img> whose tag matches one of the given | |
| 83 | - * exclusion substrings. Mirrors what Lazy_Loader does for an excluded image | |
| 84 | - * inside the_content, but page-wide so heroes outside it are covered too. | |
| 85 | - * (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. | |
| 86 | 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 | + * | |
| 87 | 162 | * @param string $html Full page HTML. |
| 88 | 163 | * @param string[] $exclusions Substring patterns identifying above-the-fold heroes. |
| 89 | 164 | */ |
| 90 | 165 | private static function promote_excluded_images( string $html, array $exclusions ): string { |
| 91 | - return (string) preg_replace_callback( | |
| 92 | - '#<img\b(?:"[^"]*"|\'[^\']*\'|[^>"\'])*>#i', | |
| 93 | - static function ( array $m ) use ( $exclusions ) { | |
| 94 | - $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]; | |
| 95 | 174 | foreach ( $exclusions as $needle ) { |
| 96 | 175 | if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { |
| 97 | 176 | $tag = (string) preg_replace( '#\s*\bloading=(["\'])\s*lazy\s*\1#i', '', $tag ); |
| 98 | - $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 | + } | |
| 99 | 181 | if ( ! preg_match( '#\bdecoding=#i', $tag ) ) { |
| 100 | 182 | $tag = (string) preg_replace( '#<img\b#i', '<img decoding="async"', $tag, 1 ); |
| 101 | 183 | } |
| 102 | 184 | return $tag; |
| @@ -103,10 +185,15 @@ | ||
| 103 | 185 | } |
| 104 | 186 | } |
| 105 | 187 | return $tag; |
| 106 | 188 | }, |
| 107 | - $html | |
| 189 | + $html, | |
| 190 | + -1, | |
| 191 | + $count, | |
| 192 | + PREG_OFFSET_CAPTURE | |
| 108 | 193 | ); |
| 194 | + | |
| 195 | + return is_string( $result ) ? $result : $html; | |
| 109 | 196 | } |
| 110 | 197 | |
| 111 | 198 | /** |
| 112 | 199 | * Build preconnect <link>s for detected font hosts + user hosts. |
| @@ -160,17 +247,58 @@ | ||
| 160 | 247 | * |
| 161 | 248 | * @param string $html Page HTML. |
| 162 | 249 | * @param int $count How many top images to preload. |
| 163 | 250 | * @param string[] $exclusions Substring patterns that exempt an <img>. |
| 251 | + * @param string $page_url The page being served, for the candidate seam. | |
| 164 | 252 | * @return array{0:string,1:string} [rewritten html, preload markup] |
| 165 | 253 | */ |
| 166 | - private static function build_lcp_preload( string $html, int $count, array $exclusions ): array { | |
| 254 | + private static function build_lcp_preload( string $html, int $count, array $exclusions, string $page_url = '' ): array { | |
| 167 | 255 | if ( $count < 1 ) { |
| 168 | 256 | return array( $html, '' ); |
| 169 | 257 | } |
| 170 | 258 | |
| 259 | + /** | |
| 260 | + * Supply the LCP preload candidate for this page instead of the | |
| 261 | + * automatic pick. | |
| 262 | + * | |
| 263 | + * The automatic pick reads the served HTML, so it cannot see a hero set | |
| 264 | + * as a background in an external stylesheet, and it skips an <img> that | |
| 265 | + * WordPress core marked lazy. An extension that has measured the page in | |
| 266 | + * a browser can answer instead. Return null to leave the automatic pick | |
| 267 | + * in charge; return an array to replace it: | |
| 268 | + * | |
| 269 | + * - `url` (string) the image to preload; an empty string means the | |
| 270 | + * page has no LCP image, so nothing is preloaded; | |
| 271 | + * - `srcset` (string, optional) and `sizes` (string, optional), emitted | |
| 272 | + * as `imagesrcset` / `imagesizes`; | |
| 273 | + * - `kind` (string, optional) `img` or `background`, passed to the URL | |
| 274 | + * filters below. An image no <img> on the page carries is a | |
| 275 | + * background regardless; | |
| 276 | + * - `media` (string, optional) a media query for the preload link. | |
| 277 | + * | |
| 278 | + * Or a list of up to three such arrays, each with its own `media`, when | |
| 279 | + * phones and desktops paint different images: each device then preloads | |
| 280 | + * its own. Any `media` (also on a single answer) sets the matched <img>s | |
| 281 | + * to `loading="lazy"`, so a device never downloads the other one's hero. | |
| 282 | + * | |
| 283 | + * Anything else (an empty list, a URL that is not a string, a URL | |
| 284 | + * esc_url() refuses, a relative URL such as `hero.png`, a `media` that | |
| 285 | + * is not a string, over 200 characters or holding `<`, `>` or `"`) is | |
| 286 | + * not understood, and the automatic pick runs. | |
| 287 | + * | |
| 288 | + * The `<img>` carrying that URL, if the page has one, gets | |
| 289 | + * `fetchpriority="high"` and loses `loading="lazy"`. The URL filters | |
| 290 | + * below (`xspeed_lcp_preload_url` and friends) still apply. | |
| 291 | + * | |
| 292 | + * @param array|null $candidate Null = no opinion. | |
| 293 | + * @param string $page_url The page being served; empty outside a request. | |
| 294 | + */ | |
| 295 | + $supplied = self::supplied_candidates( apply_filters( 'xspeed_lcp_preload_candidate', null, $page_url ) ); | |
| 296 | + if ( null !== $supplied ) { | |
| 297 | + return self::build_supplied_preload( $html, $supplied ); | |
| 298 | + } | |
| 299 | + | |
| 171 | 300 | $preload = ''; |
| 172 | - $done = 0; | |
| 173 | 301 | |
| 174 | 302 | // Snapshot of already-present preload markup, for idempotency: a second |
| 175 | 303 | // pass (e.g. cache-off ob_start over an already-processed body) must not |
| 176 | 304 | // re-emit a <link> for an image we preloaded before. |
| @@ -175,64 +303,203 @@ | ||
| 175 | 303 | // pass (e.g. cache-off ob_start over an already-processed body) must not |
| 176 | 304 | // re-emit a <link> for an image we preloaded before. |
| 177 | 305 | $existing = $html; |
| 178 | 306 | |
| 179 | - // Walk <img> tags in document order. preg_replace_callback lets us | |
| 180 | - // rewrite the tag (add fetchpriority) and harvest the src in one pass. | |
| 181 | - $html = preg_replace_callback( | |
| 182 | - '#<img\b[^>]*>#i', | |
| 183 | - static function ( array $m ) use ( &$done, &$preload, $count, $exclusions, $existing ) { | |
| 184 | - $tag = $m[0]; | |
| 185 | - if ( $done >= $count ) { | |
| 186 | - return $tag; | |
| 307 | + // PASS 1 — collect every eligible <img> and score it. | |
| 308 | + // | |
| 309 | + // This used to preload the first N eligible tags in DOCUMENT ORDER. | |
| 310 | + // Position is not a proxy for rendered size: on real pages the first | |
| 311 | + // images are header chrome, breadcrumbs or badge rows, and the actual | |
| 312 | + // LCP element is a hero further down. Preloading the wrong image gains | |
| 313 | + // nothing — it just adds a high-priority request competing with the | |
| 314 | + // one that matters, and the feature reported success either way. The | |
| 315 | + // marker list and size gate were heuristics layered on top of the | |
| 316 | + // wrong primitive rather than replacing it. (#96) | |
| 317 | + $candidates = array(); | |
| 318 | + // Markup hidden on arrival (a closed <details>, `hidden`, inline | |
| 319 | + // display:none) cannot paint as LCP either. Preloading the hero of a | |
| 320 | + // closed accordion spent the page's one High fetch on it. (#558) | |
| 321 | + $skip_ranges = array_merge( self::chrome_container_ranges( $html ), Lazy_Loader::hidden_ranges( $html ) ); | |
| 322 | + // <img> indexes that can't be the LCP however they are marked: chrome, | |
| 323 | + // hidden, or a logo/icon. Only these give up a stray `high`. (#558) | |
| 324 | + $ruled_out = array(); | |
| 325 | + if ( preg_match_all( '#<img\b[^>]*>#i', $html, $matches, PREG_OFFSET_CAPTURE ) ) { | |
| 326 | + foreach ( $matches[0] as $index => $match ) { | |
| 327 | + [ $tag, $offset ] = $match; | |
| 328 | + | |
| 329 | + // Skip anything the user excluded. | |
| 330 | + if ( self::matches_any( $tag, $exclusions ) ) { | |
| 331 | + continue; | |
| 187 | 332 | } |
| 188 | 333 | |
| 189 | - // Skip anything the user excluded. | |
| 190 | - foreach ( $exclusions as $needle ) { | |
| 191 | - if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 192 | - return $tag; | |
| 193 | - } | |
| 334 | + // An image inside <footer>/<nav>/<aside> is site chrome by | |
| 335 | + // construction — a footer brand strip or FAQ illustration can | |
| 336 | + // never be the LCP element, whatever size it declares. On the | |
| 337 | + // FBS-84576 repro these decoys outranked the real hero three | |
| 338 | + // times on one layout. | |
| 339 | + if ( self::offset_in_ranges( $offset, $skip_ranges ) || Lazy_Loader::tag_is_hidden( $tag, 'img' ) ) { | |
| 340 | + $ruled_out[ $index ] = true; | |
| 341 | + continue; | |
| 194 | 342 | } |
| 195 | 343 | |
| 344 | + // An image the theme explicitly lazy-loads is never the | |
| 345 | + // intended LCP — the author has already said "this can wait". | |
| 346 | + // Preloading it would contradict the markup and steal | |
| 347 | + // bandwidth from the image that matters. (FBS-84576) | |
| 348 | + if ( 'lazy' === strtolower( self::attr( $tag, 'loading' ) ) ) { | |
| 349 | + continue; | |
| 350 | + } | |
| 351 | + | |
| 196 | 352 | // Resolve the EFFECTIVE image URL. Page builders + JS lazy |
| 197 | 353 | // loaders park a placeholder (a data: URI or a 1px spacer) in |
| 198 | 354 | // `src` and the real URL in `data-src`, so the hero the browser |
| 199 | - // actually paints is behind data-src. Reading `src` alone made | |
| 200 | - // the LCP picker skip the real hero and fasten onto a later plain | |
| 201 | - // <img> decoy. Prefer data-src when src is a placeholder. | |
| 202 | - // (FBS-83553 H1) | |
| 355 | + // actually paints is behind data-src. (FBS-83553 H1) | |
| 203 | 356 | [ $src, $srcset, $sizes ] = self::effective_image_src( $tag ); |
| 204 | 357 | if ( '' === $src ) { |
| 205 | - return $tag; // no real URL (pure data-URI spacer, no data-src). | |
| 358 | + continue; // no real URL (pure data-URI spacer, no data-src). | |
| 206 | 359 | } |
| 207 | 360 | |
| 208 | - // Size gate: never spend the LCP preload budget on an image that | |
| 209 | - // is obviously not the hero — a logo/icon/avatar. When explicit | |
| 210 | - // width & height are on the tag and it's small in BOTH dimensions, | |
| 211 | - // skip it and keep looking. Missing dimensions → don't guess, let | |
| 212 | - // it through. (FBS-83553 H1 "logo before hero".) | |
| 361 | + // Chrome markers / explicit opt-out / obviously-tiny images | |
| 362 | + // never compete. (FBS-83553 H1 "logo before hero".) | |
| 213 | 363 | if ( self::looks_too_small( $tag ) ) { |
| 214 | - return $tag; | |
| 364 | + $ruled_out[ $index ] = true; | |
| 365 | + continue; | |
| 215 | 366 | } |
| 216 | 367 | |
| 217 | - // Idempotency: if this src is already the target of a | |
| 218 | - // rel="preload" as="image" link, count it as done (so the | |
| 219 | - // budget is respected) but don't emit a duplicate <link>. | |
| 220 | - $already = (bool) preg_match( | |
| 221 | - '#rel=["\']preload["\'][^>]*as=["\']image["\'][^>]*' . preg_quote( $src, '#' ) . '#i', | |
| 222 | - $existing | |
| 368 | + $candidates[] = array( | |
| 369 | + 'tag' => $tag, | |
| 370 | + 'src' => $src, | |
| 371 | + 'srcset' => $srcset, | |
| 372 | + 'sizes' => $sizes, | |
| 373 | + 'score' => self::weighted_score( self::lcp_score( $tag, $srcset ), $tag, $index ), | |
| 374 | + 'order' => $index, | |
| 375 | + 'offset' => $offset, | |
| 223 | 376 | ); |
| 224 | - if ( ! $already ) { | |
| 225 | - $preload .= self::preload_link( $src, $srcset, $sizes ); | |
| 377 | + } | |
| 378 | + } | |
| 379 | + | |
| 380 | + // PASS 1b — the same for CSS background images. | |
| 381 | + // | |
| 382 | + // On a page builder the hero is usually a background-image on the | |
| 383 | + // section, not an <img>, so an <img>-only candidate set never contained | |
| 384 | + // the element that actually paints as LCP. It preloaded whatever <img> | |
| 385 | + // happened to be there — measured at 0ms against the feature switched | |
| 386 | + // off, while spending a high-priority fetch on the critical path — or, | |
| 387 | + // on a page with no <img> at all, emitted nothing. (#247) | |
| 388 | + foreach ( self::background_candidates( $html, $exclusions, $skip_ranges ) as $bg ) { | |
| 389 | + $candidates[] = $bg; | |
| 390 | + } | |
| 391 | + | |
| 392 | + // PASS 1c — <video poster="…">. A full-screen hero video paints its | |
| 393 | + // poster first, and that first frame IS the LCP; measured on a live | |
| 394 | + // page, a preloaded poster cut the LCP load delay from 1.5 s to 21 ms. | |
| 395 | + foreach ( self::video_poster_candidates( $html, $exclusions, $skip_ranges ) as $vp ) { | |
| 396 | + $candidates[] = $vp; | |
| 397 | + } | |
| 398 | + | |
| 399 | + // PASS 1d — background rules in inline <style> blocks. Page builders | |
| 400 | + // put the hero's background-image in generated per-post CSS printed | |
| 401 | + // inline (Elementor's `.elementor-N .elementor-element-X` rules), not | |
| 402 | + // in a style attribute — so PASS 1b never saw the element that | |
| 403 | + // actually paints as LCP on five of seven measured sites. External | |
| 404 | + // stylesheets stay out for the same reasons as before (#247): fetching | |
| 405 | + // CSS from an output-buffer pass costs more than the preload saves. | |
| 406 | + foreach ( self::style_block_candidates( $html, $exclusions, $skip_ranges ) as $sb ) { | |
| 407 | + $candidates[] = $sb; | |
| 408 | + } | |
| 409 | + | |
| 410 | + // A background video with no poster, ahead of every candidate, is the | |
| 411 | + // hero. Nothing in its box is preloadable, and every image after it | |
| 412 | + // sits lower on the page. Preloading the first of those spent the one | |
| 413 | + // high-priority fetch on an image below the fold, ahead of the CSS. | |
| 414 | + // Those images also give up a stray `high` from the lazy pass. | |
| 415 | + $video_at = self::background_video_offset( $html, $skip_ranges ); | |
| 416 | + if ( null !== $video_at ) { | |
| 417 | + foreach ( $candidates as $i => $c ) { | |
| 418 | + if ( $c['offset'] > $video_at ) { | |
| 419 | + if ( empty( $c['background'] ) ) { | |
| 420 | + $ruled_out[ $c['order'] ] = true; | |
| 421 | + } | |
| 422 | + unset( $candidates[ $i ] ); | |
| 226 | 423 | } |
| 227 | - $done++; | |
| 424 | + } | |
| 425 | + } | |
| 228 | 426 | |
| 427 | + if ( empty( $candidates ) ) { | |
| 428 | + if ( null !== $video_at ) { | |
| 429 | + $html = self::demote_images( $html, $ruled_out ); | |
| 430 | + } | |
| 431 | + return array( $html, '' ); | |
| 432 | + } | |
| 433 | + | |
| 434 | + // Rank by score, biggest first. Document order breaks ties, so two | |
| 435 | + // equally-sized images (or two of unknown size) keep the previous | |
| 436 | + // first-wins behaviour — the change only matters when we can actually | |
| 437 | + // tell one is larger. | |
| 438 | + usort( | |
| 439 | + $candidates, | |
| 440 | + static function ( array $a, array $b ) { | |
| 441 | + if ( $a['score'] === $b['score'] ) { | |
| 442 | + return $a['order'] <=> $b['order']; | |
| 443 | + } | |
| 444 | + return $b['score'] <=> $a['score']; | |
| 445 | + } | |
| 446 | + ); | |
| 447 | + | |
| 448 | + $winners = array_slice( $candidates, 0, $count ); | |
| 449 | + | |
| 450 | + // PASS 2 — emit the preload links and promote the winning tags. | |
| 451 | + $chosen = array(); | |
| 452 | + foreach ( $winners as $w ) { | |
| 453 | + // Idempotency: if this src is already the target of a | |
| 454 | + // rel="preload" as="image" link, still promote the tag but don't | |
| 455 | + // emit a duplicate <link>. | |
| 456 | + $already = (bool) preg_match( | |
| 457 | + '#rel=["\']preload["\'][^>]*as=["\']image["\'][^>]*' . preg_quote( $w['src'], '#' ) . '#i', | |
| 458 | + $existing | |
| 459 | + ); | |
| 460 | + if ( ! $already ) { | |
| 461 | + $preload .= self::preload_link( $w['src'], $w['srcset'], $w['sizes'], empty( $w['background'] ) ? 'img' : 'background' ); | |
| 462 | + } | |
| 463 | + // Only <img> winners are promoted in PASS 2 — there is no | |
| 464 | + // fetchpriority/loading attribute to fix on a background element, | |
| 465 | + // and its `order` is offset past every <img> index precisely so it | |
| 466 | + // can never select one for rewriting. | |
| 467 | + if ( empty( $w['background'] ) ) { | |
| 468 | + $chosen[ $w['order'] ] = true; | |
| 469 | + } | |
| 470 | + } | |
| 471 | + | |
| 472 | + // Rewrite only the winning tags. Counting occurrences rather than | |
| 473 | + // matching on tag text, because the same markup can legitimately | |
| 474 | + // appear more than once on a page and only the ranked instance should | |
| 475 | + // be promoted. | |
| 476 | + // | |
| 477 | + // When an <img> wins, it is the page's one High image, so an image | |
| 478 | + // that can't be the LCP (a logo, an icon, a hidden panel, chrome) | |
| 479 | + // gives up any `high` it carries. That is usually the lazy pass handing | |
| 480 | + // its slot to the first image in the_content. An image that merely scored | |
| 481 | + // lower, or that the user kept out of the pick, keeps its hint: the | |
| 482 | + // ranking can be wrong, and core or the theme may have named the real | |
| 483 | + // hero. (#558) | |
| 484 | + $demote = ! empty( $chosen ) || null !== $video_at ? $ruled_out : array(); | |
| 485 | + $seen = -1; | |
| 486 | + $html = preg_replace_callback( | |
| 487 | + '#<img\b[^>]*>#i', | |
| 488 | + static function ( array $m ) use ( &$seen, $chosen, $demote ) { | |
| 489 | + ++$seen; | |
| 490 | + if ( ! isset( $chosen[ $seen ] ) ) { | |
| 491 | + if ( isset( $demote[ $seen ] ) && 'high' === strtolower( self::attr( $m[0], 'fetchpriority' ) ) ) { | |
| 492 | + return (string) preg_replace( '#\s*(?<![-\w])fetchpriority\s*=\s*(["\']?)high\1#i', '', $m[0], 1 ); | |
| 493 | + } | |
| 494 | + return $m[0]; | |
| 495 | + } | |
| 229 | 496 | // Add fetchpriority="high" AND remove any loading="lazy" the |
| 230 | 497 | // theme / WP core left on the LCP image. fetchpriority="high" |
| 231 | 498 | // with loading="lazy" is contradictory — the browser can still |
| 232 | 499 | // defer a lazy image, so preloading it while it stays lazy wins |
| 233 | 500 | // nothing. Stripping lazy is what actually lets the preload land. |
| 234 | - return self::promote_lcp_img( $tag ); | |
| 501 | + return self::promote_lcp_img( $m[0] ); | |
| 235 | 502 | }, |
| 236 | 503 | $html |
| 237 | 504 | ); |
| 238 | 505 | |
| @@ -239,8 +506,638 @@ | ||
| 239 | 506 | return array( (string) $html, $preload ); |
| 240 | 507 | } |
| 241 | 508 | |
| 242 | 509 | /** |
| 510 | + * Byte offset of the first visible background video with no poster, or | |
| 511 | + * null when the page has none. | |
| 512 | + * | |
| 513 | + * @param string $html Full page HTML. | |
| 514 | + * @param array<int,array{0:int,1:int}> $skip_ranges Chrome and hidden spans. | |
| 515 | + */ | |
| 516 | + private static function background_video_offset( string $html, array $skip_ranges ): ?int { | |
| 517 | + $body = stripos( $html, '<body' ); | |
| 518 | + if ( ! preg_match_all( '#<video\b[^>]*>#i', $html, $m, PREG_OFFSET_CAPTURE, false === $body ? 0 : $body ) ) { | |
| 519 | + return null; | |
| 520 | + } | |
| 521 | + foreach ( $m[0] as [ $tag, $offset ] ) { | |
| 522 | + if ( self::offset_in_ranges( $offset, $skip_ranges ) || Lazy_Loader::tag_is_hidden( $tag, 'video' ) ) { | |
| 523 | + continue; | |
| 524 | + } | |
| 525 | + if ( Lazy_Loader::is_background_video_without_poster( $tag ) ) { | |
| 526 | + return $offset; | |
| 527 | + } | |
| 528 | + } | |
| 529 | + return null; | |
| 530 | + } | |
| 531 | + | |
| 532 | + /** | |
| 533 | + * Strip fetchpriority="high" from the <img> tags at the given indexes. | |
| 534 | + * | |
| 535 | + * @param string $html Page HTML. | |
| 536 | + * @param array<int,bool> $indexes <img> indexes, in document order. | |
| 537 | + */ | |
| 538 | + private static function demote_images( string $html, array $indexes ): string { | |
| 539 | + if ( empty( $indexes ) ) { | |
| 540 | + return $html; | |
| 541 | + } | |
| 542 | + $seen = -1; | |
| 543 | + $out = preg_replace_callback( | |
| 544 | + '#<img\b[^>]*>#i', | |
| 545 | + static function ( array $m ) use ( &$seen, $indexes ) { | |
| 546 | + ++$seen; | |
| 547 | + if ( isset( $indexes[ $seen ] ) && 'high' === strtolower( self::attr( $m[0], 'fetchpriority' ) ) ) { | |
| 548 | + return (string) preg_replace( '#\s*(?<![-\w])fetchpriority\s*=\s*(["\']?)high\1#i', '', $m[0], 1 ); | |
| 549 | + } | |
| 550 | + return $m[0]; | |
| 551 | + }, | |
| 552 | + $html | |
| 553 | + ); | |
| 554 | + return null === $out ? $html : $out; | |
| 555 | + } | |
| 556 | + | |
| 557 | + /** | |
| 558 | + * Collect CSS `background-image` heroes as LCP candidates. | |
| 559 | + * | |
| 560 | + * Only INLINE `style` attributes are read. A background declared in an | |
| 561 | + * external stylesheet is invisible here by design: resolving it would mean | |
| 562 | + * fetching and parsing CSS from inside an output-buffer pass, and the URL a | |
| 563 | + * selector resolves to depends on cascade order we cannot evaluate from | |
| 564 | + * markup. Builders that put the hero in a generated per-post stylesheet are | |
| 565 | + * therefore still unserved — worth doing, but not at this cost. (#247) | |
| 566 | + * | |
| 567 | + * Scores are the element's declared pixel area so a background competes | |
| 568 | + * against an <img> in the SAME units — the whole point being that the | |
| 569 | + * bigger of the two should win regardless of which kind it is. | |
| 570 | + * | |
| 571 | + * @param string $html Full page HTML. | |
| 572 | + * @param string[] $exclusions Substring patterns the user excluded. | |
| 573 | + * @param array<int,array{0:int,1:int}> $skip_ranges Byte ranges of chrome containers. | |
| 574 | + * @return array<int,array{tag:string,src:string,srcset:string,sizes:string,score:float,order:int,background:bool}> | |
| 575 | + */ | |
| 576 | + private static function background_candidates( string $html, array $exclusions, array $skip_ranges ): array { | |
| 577 | + 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 ) ) { | |
| 578 | + return array(); | |
| 579 | + } | |
| 580 | + | |
| 581 | + $found = array(); | |
| 582 | + foreach ( $matches[0] as $index => $match ) { | |
| 583 | + [ $tag, $offset ] = $match; | |
| 584 | + | |
| 585 | + // The same chrome-container gate as <img>: a background painted | |
| 586 | + // inside <footer>/<nav>/<aside> is never the hero. (FBS-84576) | |
| 587 | + if ( self::offset_in_ranges( $offset, $skip_ranges ) ) { | |
| 588 | + continue; | |
| 589 | + } | |
| 590 | + | |
| 591 | + $style = self::attr( $tag, 'style' ); | |
| 592 | + if ( '' === $style || false === stripos( $style, 'background' ) ) { | |
| 593 | + continue; | |
| 594 | + } | |
| 595 | + | |
| 596 | + $src = self::background_url( $style ); | |
| 597 | + if ( '' === $src ) { | |
| 598 | + continue; | |
| 599 | + } | |
| 600 | + | |
| 601 | + foreach ( $exclusions as $needle ) { | |
| 602 | + if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 603 | + continue 2; | |
| 604 | + } | |
| 605 | + } | |
| 606 | + | |
| 607 | + // Same chrome/opt-out gates as <img>. A logo painted as a background | |
| 608 | + // is no more the hero than a logo in an <img>. | |
| 609 | + if ( self::looks_too_small( $tag ) ) { | |
| 610 | + continue; | |
| 611 | + } | |
| 612 | + | |
| 613 | + $area = self::style_area( $style ); | |
| 614 | + if ( 0 === $area ) { | |
| 615 | + // Nothing readable. Deliberately non-zero for the same reason | |
| 616 | + // UNKNOWN_SIZE_SCORE is: an unmeasurable background must still | |
| 617 | + // beat nothing on a page that declares no sizes at all, while | |
| 618 | + // losing to anything we can actually measure. | |
| 619 | + $area = self::UNKNOWN_SIZE_SCORE; | |
| 620 | + } | |
| 621 | + | |
| 622 | + $found[] = array( | |
| 623 | + 'tag' => $tag, | |
| 624 | + 'src' => $src, | |
| 625 | + 'srcset' => '', | |
| 626 | + 'sizes' => '', | |
| 627 | + 'score' => (float) $area, | |
| 628 | + // Offset so a background never ties ahead of an <img> that | |
| 629 | + // appeared earlier in the document; ties still break on order. | |
| 630 | + 'order' => 100000 + $index, | |
| 631 | + 'offset' => $offset, | |
| 632 | + 'background' => true, | |
| 633 | + ); | |
| 634 | + } | |
| 635 | + | |
| 636 | + return $found; | |
| 637 | + } | |
| 638 | + | |
| 639 | + /** | |
| 640 | + * Collect `<video poster="…">` first frames as LCP candidates. | |
| 641 | + * | |
| 642 | + * The poster is what the viewer sees until (and unless) the video plays — | |
| 643 | + * on a background-video hero, delayed by the Lazy module, it is the ONLY | |
| 644 | + * frame the initial paint has. Scored like a background: the element's | |
| 645 | + * declared inline-style area, or the unknown-size floor, with the order | |
| 646 | + * offset past every <img> so a poster never ties ahead of one. | |
| 647 | + * | |
| 648 | + * @param string $html Full page HTML. | |
| 649 | + * @param string[] $exclusions Substring patterns the user excluded. | |
| 650 | + * @param array<int,array{0:int,1:int}> $skip_ranges Byte ranges of chrome containers. | |
| 651 | + * @return array<int,array{tag:string,src:string,srcset:string,sizes:string,score:float,order:int,background:bool}> | |
| 652 | + */ | |
| 653 | + private static function video_poster_candidates( string $html, array $exclusions, array $skip_ranges ): array { | |
| 654 | + if ( ! preg_match_all( '#<video\b[^>]*\bposter\s*=\s*(["\'])(.*?)\1[^>]*>#i', $html, $matches, PREG_OFFSET_CAPTURE ) ) { | |
| 655 | + return array(); | |
| 656 | + } | |
| 657 | + | |
| 658 | + $found = array(); | |
| 659 | + foreach ( $matches[0] as $index => $match ) { | |
| 660 | + [ $tag, $offset ] = $match; | |
| 661 | + $src = trim( html_entity_decode( $matches[2][ $index ][0], ENT_QUOTES ) ); | |
| 662 | + if ( '' === $src || 0 === stripos( $src, 'data:' ) ) { | |
| 663 | + continue; | |
| 664 | + } | |
| 665 | + if ( self::offset_in_ranges( $offset, $skip_ranges ) ) { | |
| 666 | + continue; | |
| 667 | + } | |
| 668 | + foreach ( $exclusions as $needle ) { | |
| 669 | + if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 670 | + continue 2; | |
| 671 | + } | |
| 672 | + } | |
| 673 | + if ( self::looks_too_small( $tag ) ) { | |
| 674 | + continue; | |
| 675 | + } | |
| 676 | + | |
| 677 | + $area = self::style_area( self::attr( $tag, 'style' ) ); | |
| 678 | + if ( 0 === $area ) { | |
| 679 | + $area = self::UNKNOWN_SIZE_SCORE; | |
| 680 | + } | |
| 681 | + | |
| 682 | + $found[] = array( | |
| 683 | + 'tag' => $tag, | |
| 684 | + 'src' => $src, | |
| 685 | + 'srcset' => '', | |
| 686 | + 'sizes' => '', | |
| 687 | + 'score' => (float) $area, | |
| 688 | + 'order' => 100000 + $index, | |
| 689 | + 'offset' => $offset, | |
| 690 | + 'background' => true, | |
| 691 | + ); | |
| 692 | + } | |
| 693 | + | |
| 694 | + return $found; | |
| 695 | + } | |
| 696 | + | |
| 697 | + /** | |
| 698 | + * How many <style>-block background rules are considered per page. The | |
| 699 | + * scan is linear, but each rule costs one class-lookup pass over the | |
| 700 | + * body, so a pathological page (thousands of generated rules) is capped | |
| 701 | + * rather than trusted. | |
| 702 | + */ | |
| 703 | + private const STYLE_RULE_BUDGET = 40; | |
| 704 | + | |
| 705 | + /** | |
| 706 | + * Collect background-image rules from inline <style> blocks whose | |
| 707 | + * selector matches an element in the body. | |
| 708 | + * | |
| 709 | + * The match is deliberately narrow: the rule's RIGHTMOST simple selector | |
| 710 | + * must carry a class or id, and the first element in the body bearing it | |
| 711 | + * (outside chrome containers) is taken as the painted element. Rules | |
| 712 | + * inside @media (or any other at-rule block) are skipped — a desktop-only | |
| 713 | + * background preloaded on mobile is a wasted high-priority fetch, and the | |
| 714 | + * markup gives no viewport to resolve the query against. | |
| 715 | + * | |
| 716 | + * @param string $html Full page HTML. | |
| 717 | + * @param string[] $exclusions Substring patterns the user excluded. | |
| 718 | + * @param array<int,array{0:int,1:int}> $skip_ranges Byte ranges of chrome containers. | |
| 719 | + * @return array<int,array{tag:string,src:string,srcset:string,sizes:string,score:float,order:int,background:bool}> | |
| 720 | + */ | |
| 721 | + private static function style_block_candidates( string $html, array $exclusions, array $skip_ranges ): array { | |
| 722 | + if ( ! preg_match_all( '#<style\b[^>]*>(.*?)</style\s*>#is', $html, $blocks ) ) { | |
| 723 | + return array(); | |
| 724 | + } | |
| 725 | + | |
| 726 | + $found = array(); | |
| 727 | + $budget = self::STYLE_RULE_BUDGET; | |
| 728 | + foreach ( $blocks[1] as $css ) { | |
| 729 | + if ( $budget <= 0 ) { | |
| 730 | + break; | |
| 731 | + } | |
| 732 | + $css = self::strip_at_rule_blocks( $css ); | |
| 733 | + if ( false === stripos( $css, 'url(' ) ) { | |
| 734 | + continue; | |
| 735 | + } | |
| 736 | + // One flat rule at a time: selector list up to '{', body to '}'. | |
| 737 | + if ( ! preg_match_all( '#(?:^|})\s*([^{}]{1,512})\{([^{}]*)\}#s', $css, $rules, PREG_SET_ORDER ) ) { | |
| 738 | + continue; | |
| 739 | + } | |
| 740 | + foreach ( $rules as $rule ) { | |
| 741 | + if ( $budget <= 0 ) { | |
| 742 | + break 2; | |
| 743 | + } | |
| 744 | + if ( false === stripos( $rule[2], 'url(' ) ) { | |
| 745 | + continue; | |
| 746 | + } | |
| 747 | + $src = self::background_url( $rule[2] ); | |
| 748 | + if ( '' === $src ) { | |
| 749 | + continue; | |
| 750 | + } | |
| 751 | + --$budget; | |
| 752 | + // First selector of the list, rightmost compound of it. | |
| 753 | + $selector = trim( (string) strtok( $rule[1], ',' ) ); | |
| 754 | + $parts = preg_split( '#[\s>+~]+#', $selector ); | |
| 755 | + $last = (string) end( $parts ); | |
| 756 | + // The last class or id token of that compound. Pseudo-classes | |
| 757 | + // (:hover, ::before) mean the background is not the initial | |
| 758 | + // paint, so they disqualify the rule. | |
| 759 | + if ( false !== strpos( $last, ':' ) ) { | |
| 760 | + continue; | |
| 761 | + } | |
| 762 | + if ( ! preg_match( '#([.\#])([-\w]+)$#', $last, $tok ) ) { | |
| 763 | + continue; | |
| 764 | + } | |
| 765 | + $el = '.' === $tok[1] | |
| 766 | + ? self::first_element_with_class( $html, $tok[2], $skip_ranges ) | |
| 767 | + : self::first_element_with_id( $html, $tok[2], $skip_ranges ); | |
| 768 | + if ( null === $el ) { | |
| 769 | + continue; | |
| 770 | + } | |
| 771 | + [ $tag, $offset ] = $el; | |
| 772 | + foreach ( $exclusions as $needle ) { | |
| 773 | + if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 774 | + continue 2; | |
| 775 | + } | |
| 776 | + } | |
| 777 | + if ( self::looks_too_small( $tag ) ) { | |
| 778 | + continue; | |
| 779 | + } | |
| 780 | + $area = self::style_area( self::attr( $tag, 'style' ) ); | |
| 781 | + if ( 0 === $area ) { | |
| 782 | + $area = self::UNKNOWN_SIZE_SCORE; | |
| 783 | + } | |
| 784 | + $found[] = array( | |
| 785 | + 'tag' => $tag, | |
| 786 | + 'src' => $src, | |
| 787 | + 'srcset' => '', | |
| 788 | + 'sizes' => '', | |
| 789 | + 'score' => (float) $area, | |
| 790 | + // Offset past the inline-style backgrounds: a rule-matched | |
| 791 | + // background is one inference step less certain, so it must | |
| 792 | + // never tie ahead of one read straight off the element. | |
| 793 | + 'order' => 200000 + $offset, | |
| 794 | + 'offset' => $offset, | |
| 795 | + 'background' => true, | |
| 796 | + ); | |
| 797 | + } | |
| 798 | + } | |
| 799 | + | |
| 800 | + return $found; | |
| 801 | + } | |
| 802 | + | |
| 803 | + /** | |
| 804 | + * CSS with every at-rule BLOCK (@media, @supports, @container, …) removed, | |
| 805 | + * by brace depth — a regex cannot pair nested braces. Flat at-rules | |
| 806 | + * (@import, @charset) have no block and pass through harmlessly. | |
| 807 | + */ | |
| 808 | + private static function strip_at_rule_blocks( string $css ): string { | |
| 809 | + $out = ''; | |
| 810 | + $len = strlen( $css ); | |
| 811 | + $i = 0; | |
| 812 | + while ( $i < $len ) { | |
| 813 | + $at = strpos( $css, '@', $i ); | |
| 814 | + if ( false === $at ) { | |
| 815 | + return $out . substr( $css, $i ); | |
| 816 | + } | |
| 817 | + $brace = strpos( $css, '{', $at ); | |
| 818 | + $semi = strpos( $css, ';', $at ); | |
| 819 | + $out .= substr( $css, $i, $at - $i ); | |
| 820 | + if ( false === $brace || ( false !== $semi && $semi < $brace ) ) { | |
| 821 | + // Flat at-rule — skip to its semicolon (or end). | |
| 822 | + $i = false === $semi ? $len : $semi + 1; | |
| 823 | + continue; | |
| 824 | + } | |
| 825 | + // Block at-rule — skip to its matching close brace. | |
| 826 | + $depth = 1; | |
| 827 | + $i = $brace + 1; | |
| 828 | + while ( $i < $len && $depth > 0 ) { | |
| 829 | + $c = $css[ $i ]; | |
| 830 | + if ( '{' === $c ) { | |
| 831 | + ++$depth; | |
| 832 | + } elseif ( '}' === $c ) { | |
| 833 | + --$depth; | |
| 834 | + } | |
| 835 | + ++$i; | |
| 836 | + } | |
| 837 | + } | |
| 838 | + return $out; | |
| 839 | + } | |
| 840 | + | |
| 841 | + /** | |
| 842 | + * The first element in the BODY carrying $class (outside chrome ranges), | |
| 843 | + * as [tag, offset], or null. Body-only, so a head <meta> can never match | |
| 844 | + * and a hit's offset is comparable with the chrome ranges. | |
| 845 | + * | |
| 846 | + * @return array{0:string,1:int}|null | |
| 847 | + */ | |
| 848 | + private static function first_element_with_class( string $html, string $class, array $skip_ranges ): ?array { | |
| 849 | + $body = stripos( $html, '<body' ); | |
| 850 | + $from = false === $body ? 0 : $body; | |
| 851 | + if ( ! preg_match_all( '#<[a-z][^>]*\bclass\s*=\s*(["\'])[^"\']*(?<![-\w])' . preg_quote( $class, '#' ) . '(?![-\w])[^"\']*\1[^>]*>#i', $html, $m, PREG_OFFSET_CAPTURE, $from ) ) { | |
| 852 | + return null; | |
| 853 | + } | |
| 854 | + foreach ( $m[0] as $match ) { | |
| 855 | + if ( ! self::offset_in_ranges( $match[1], $skip_ranges ) ) { | |
| 856 | + return array( $match[0], $match[1] ); | |
| 857 | + } | |
| 858 | + } | |
| 859 | + return null; | |
| 860 | + } | |
| 861 | + | |
| 862 | + /** | |
| 863 | + * The first element carrying id="$id" (outside chrome ranges), as | |
| 864 | + * [tag, offset], or null. | |
| 865 | + * | |
| 866 | + * @return array{0:string,1:int}|null | |
| 867 | + */ | |
| 868 | + private static function first_element_with_id( string $html, string $id, array $skip_ranges ): ?array { | |
| 869 | + $body = stripos( $html, '<body' ); | |
| 870 | + $from = false === $body ? 0 : $body; | |
| 871 | + if ( ! preg_match( '#<[a-z][^>]*\bid\s*=\s*(["\'])' . preg_quote( $id, '#' ) . '\1[^>]*>#i', $html, $m, PREG_OFFSET_CAPTURE, $from ) ) { | |
| 872 | + return null; | |
| 873 | + } | |
| 874 | + if ( self::offset_in_ranges( $m[0][1], $skip_ranges ) ) { | |
| 875 | + return null; | |
| 876 | + } | |
| 877 | + return array( $m[0][0], $m[0][1] ); | |
| 878 | + } | |
| 879 | + | |
| 880 | + /** | |
| 881 | + * Pull a real image URL out of a `background`/`background-image` declaration. | |
| 882 | + * | |
| 883 | + * Returns '' for anything with nothing to fetch: a gradient (which is a | |
| 884 | + * background-image but not a resource), a data: URI, or `none`. | |
| 885 | + */ | |
| 886 | + private static function background_url( string $style ): string { | |
| 887 | + // Decode BEFORE parsing. Builders emit the url() quotes HTML-encoded | |
| 888 | + // inside a style attribute (url("/hero.jpg")), and `"` | |
| 889 | + // carries a semicolon — so splitting the declaration on `;` first | |
| 890 | + // truncated the value to `url("` and found no URL at all. | |
| 891 | + $style = html_entity_decode( $style, ENT_QUOTES ); | |
| 892 | + | |
| 893 | + if ( ! preg_match( '#background(?:-image)?\s*:\s*((?:[^;\'"]|"[^"]*"|\'[^\']*\')+)#i', $style, $decl ) ) { | |
| 894 | + return ''; | |
| 895 | + } | |
| 896 | + if ( ! preg_match( '#url\(\s*(["\']?)(.*?)\1\s*\)#is', $decl[1], $m ) ) { | |
| 897 | + return ''; | |
| 898 | + } | |
| 899 | + $url = trim( $m[2] ); | |
| 900 | + if ( '' === $url || 0 === stripos( $url, 'data:' ) ) { | |
| 901 | + return ''; | |
| 902 | + } | |
| 903 | + return $url; | |
| 904 | + } | |
| 905 | + | |
| 906 | + /** | |
| 907 | + * Declared pixel area from an inline style, or 0 when it can't be read. | |
| 908 | + * | |
| 909 | + * Only px is honoured. A percentage or viewport unit resolves against a | |
| 910 | + * containing block we cannot see from markup, and guessing one produced the | |
| 911 | + * wrong winner more often than declining to. | |
| 912 | + */ | |
| 913 | + private static function style_area( string $style ): int { | |
| 914 | + $w = self::style_px( $style, 'width' ); | |
| 915 | + $h = self::style_px( $style, 'height' ); | |
| 916 | + if ( $w > 0 && $h > 0 ) { | |
| 917 | + return $w * $h; | |
| 918 | + } | |
| 919 | + if ( $w > 0 ) { | |
| 920 | + return (int) round( $w * $w * self::ASSUMED_ASPECT_RATIO ); | |
| 921 | + } | |
| 922 | + return 0; | |
| 923 | + } | |
| 924 | + | |
| 925 | + /** One px-valued CSS length from an inline style, or 0. */ | |
| 926 | + private static function style_px( string $style, string $prop ): int { | |
| 927 | + if ( preg_match( '#(?:^|;)\s*' . preg_quote( $prop, '#' ) . '\s*:\s*(\d+(?:\.\d+)?)px#i', $style, $m ) ) { | |
| 928 | + return (int) round( (float) $m[1] ); | |
| 929 | + } | |
| 930 | + return 0; | |
| 931 | + } | |
| 932 | + | |
| 933 | + /** | |
| 934 | + * How likely is this <img> to be the LCP element? Higher wins. | |
| 935 | + * | |
| 936 | + * Rendered area is the best available proxy, and we can only read what the | |
| 937 | + * markup declares: | |
| 938 | + * | |
| 939 | + * 1. `width` × `height` attributes — the real area, when present. | |
| 940 | + * 2. The largest `srcset` / `data-srcset` candidate width, squared into a | |
| 941 | + * pseudo-area. A responsive hero usually omits width/height but ships | |
| 942 | + * a 1600w+ candidate, which says more about its size than its | |
| 943 | + * position ever did. | |
| 944 | + * 3. Nothing readable → a neutral score, so the image still competes | |
| 945 | + * (matching looks_too_small()'s "don't guess" rule) but loses to any | |
| 946 | + * image we CAN measure as larger. | |
| 947 | + * | |
| 948 | + * @param string $tag The full <img> tag. | |
| 949 | + * @param string $srcset Resolved srcset (may come from data-srcset). | |
| 950 | + */ | |
| 951 | + private static function lcp_score( string $tag, string $srcset ): float { | |
| 952 | + $w = self::attr( $tag, 'width' ); | |
| 953 | + $h = self::attr( $tag, 'height' ); | |
| 954 | + if ( '' !== $w && '' !== $h && is_numeric( $w ) && is_numeric( $h ) ) { | |
| 955 | + return (float) ( (int) $w * (int) $h ); | |
| 956 | + } | |
| 957 | + | |
| 958 | + $widest = self::widest_srcset_width( $srcset ); | |
| 959 | + if ( $widest > 0 ) { | |
| 960 | + // Estimate an AREA, not a square. Squaring the width compared a | |
| 961 | + // pseudo-area against a real one and overstated the width-only | |
| 962 | + // candidate by roughly the inverse of its aspect ratio, so a | |
| 963 | + // 1024w sidebar thumbnail (1 048 576) beat a declared 1200×600 | |
| 964 | + // hero (720 000) — a regression on exactly the mixed pages that | |
| 965 | + // document order used to get right, since the hero usually comes | |
| 966 | + // first. Assuming a 16:9 box keeps both sides in the same units. | |
| 967 | + return round( $widest * $widest * self::ASSUMED_ASPECT_RATIO ); | |
| 968 | + } | |
| 969 | + | |
| 970 | + return (float) self::UNKNOWN_SIZE_SCORE; | |
| 971 | + } | |
| 972 | + | |
| 973 | + /** | |
| 974 | + * Fold the author's own priority signals and document position into an | |
| 975 | + * area score. (FBS-84576) | |
| 976 | + * | |
| 977 | + * Boosts are ADDITIVE, in area units, so they can rescue an image whose | |
| 978 | + * size the markup doesn't declare: a hero with no width/height and no | |
| 979 | + * `w`-descriptor srcset scores UNKNOWN_SIZE_SCORE, and multiplying that | |
| 980 | + * by any factor still loses to a 548×136 logo that declares itself. This | |
| 981 | + * is exactly the live miss — the real hero carried loading="eager" | |
| 982 | + * fetchpriority="high" and lost to three dimension-declaring decoys. | |
| 983 | + * | |
| 984 | + * - fetchpriority="high" is the strongest signal there is: the author | |
| 985 | + * (or WP core's own LCP detection) has already named this image the | |
| 986 | + * hero. Worth a hero-sized area. | |
| 987 | + * - An EXPLICIT loading="eager" is a weaker but deliberate "load me | |
| 988 | + * now" (the default is eager, so writing it out is a choice). | |
| 989 | + * | |
| 990 | + * Position is a light multiplicative weight — earlier is better, but the | |
| 991 | + * spread is capped well under 5× so it can only break near-ties, never | |
| 992 | + * outrank a genuinely larger image further down (the logo-vs-hero case). | |
| 993 | + * | |
| 994 | + * @param float $score Base area score from lcp_score() / style_area(). | |
| 995 | + * @param string $tag The candidate's full tag (for the signal attrs). | |
| 996 | + * @param int $order Document-order index of the candidate. | |
| 997 | + */ | |
| 998 | + private static function weighted_score( float $score, string $tag, int $order ): float { | |
| 999 | + if ( 'high' === strtolower( self::attr( $tag, 'fetchpriority' ) ) ) { | |
| 1000 | + $score += self::FETCHPRIORITY_HIGH_BOOST; | |
| 1001 | + } | |
| 1002 | + if ( 'eager' === strtolower( self::attr( $tag, 'loading' ) ) ) { | |
| 1003 | + $score += self::EAGER_BOOST; | |
| 1004 | + } | |
| 1005 | + return $score * self::position_weight( $order ); | |
| 1006 | + } | |
| 1007 | + | |
| 1008 | + /** | |
| 1009 | + * Document-position weight: 1.25 for the first image, easing to 1.0 by | |
| 1010 | + * the tenth. The whole spread is 25%, far under the 5× area difference it | |
| 1011 | + * must never override — it exists only to keep the old first-wins | |
| 1012 | + * behaviour for images we can't tell apart. | |
| 1013 | + */ | |
| 1014 | + private static function position_weight( int $order ): float { | |
| 1015 | + return 1.0 + 0.25 * max( 0.0, 1.0 - $order / 10 ); | |
| 1016 | + } | |
| 1017 | + | |
| 1018 | + /** | |
| 1019 | + * Area-unit boost for fetchpriority="high" — roughly a 940×530 hero, so | |
| 1020 | + * an explicitly-marked image outranks any mid-page decoy even when its | |
| 1021 | + * own size is unreadable, while a genuinely huge unmarked image can still | |
| 1022 | + * beat a marked small one. | |
| 1023 | + */ | |
| 1024 | + private const FETCHPRIORITY_HIGH_BOOST = 500000.0; | |
| 1025 | + | |
| 1026 | + /** | |
| 1027 | + * Area-unit boost for an explicit loading="eager" — roughly 420×240, | |
| 1028 | + * enough to break ties in favour of the author's intent without letting | |
| 1029 | + * an eager logo outrank a plain hero. | |
| 1030 | + */ | |
| 1031 | + private const EAGER_BOOST = 100000.0; | |
| 1032 | + | |
| 1033 | + /** | |
| 1034 | + * Byte ranges of <footer>/<nav>/<aside> regions. Nesting-aware per tag | |
| 1035 | + * name (a nav inside a nav extends the range); an unclosed open tag | |
| 1036 | + * poisons through to the end of the document, which errs on the side of | |
| 1037 | + * not preloading — the safe direction, since a wrong preload is worse | |
| 1038 | + * than none. (FBS-84576) | |
| 1039 | + * | |
| 1040 | + * @return array<int,array{0:int,1:int}> [start, end] byte offsets. | |
| 1041 | + */ | |
| 1042 | + private static function chrome_container_ranges( string $html ): array { | |
| 1043 | + $ranges = array(); | |
| 1044 | + foreach ( array( 'footer', 'nav', 'aside' ) as $name ) { | |
| 1045 | + // (?=[\s/>]) rather than \b: a word boundary sits before the `-` | |
| 1046 | + // of a custom element, so `<nav\b` would swallow `<nav-menu>`. | |
| 1047 | + if ( ! preg_match_all( '#<(/?)' . $name . '(?=[\s/>])[^>]*>#i', $html, $m, PREG_OFFSET_CAPTURE ) ) { | |
| 1048 | + continue; | |
| 1049 | + } | |
| 1050 | + $depth = 0; | |
| 1051 | + $start = 0; | |
| 1052 | + foreach ( $m[0] as $i => $match ) { | |
| 1053 | + $closing = '' !== $m[1][ $i ][0]; | |
| 1054 | + if ( ! $closing ) { | |
| 1055 | + if ( 0 === $depth ) { | |
| 1056 | + $start = $match[1]; | |
| 1057 | + } | |
| 1058 | + ++$depth; | |
| 1059 | + } elseif ( $depth > 0 ) { | |
| 1060 | + --$depth; | |
| 1061 | + if ( 0 === $depth ) { | |
| 1062 | + $ranges[] = array( $start, $match[1] ); | |
| 1063 | + } | |
| 1064 | + } | |
| 1065 | + } | |
| 1066 | + if ( $depth > 0 ) { | |
| 1067 | + $ranges[] = array( $start, strlen( $html ) ); | |
| 1068 | + } | |
| 1069 | + } | |
| 1070 | + return $ranges; | |
| 1071 | + } | |
| 1072 | + | |
| 1073 | + /** | |
| 1074 | + * Does a tag contain any of the given substring patterns? | |
| 1075 | + * | |
| 1076 | + * @param string[] $patterns | |
| 1077 | + */ | |
| 1078 | + private static function matches_any( string $tag, array $patterns ): bool { | |
| 1079 | + foreach ( $patterns as $needle ) { | |
| 1080 | + if ( '' !== $needle && false !== stripos( $tag, $needle ) ) { | |
| 1081 | + return true; | |
| 1082 | + } | |
| 1083 | + } | |
| 1084 | + return false; | |
| 1085 | + } | |
| 1086 | + | |
| 1087 | + /** Does a byte offset fall inside any of the given [start, end] ranges? */ | |
| 1088 | + private static function offset_in_ranges( int $offset, array $ranges ): bool { | |
| 1089 | + foreach ( $ranges as $range ) { | |
| 1090 | + // >= on the start: a closed <details> span from | |
| 1091 | + // Lazy_Loader::hidden_ranges() opens right at the first image | |
| 1092 | + // after its <summary>. | |
| 1093 | + if ( $offset >= $range[0] && $offset < $range[1] ) { | |
| 1094 | + return true; | |
| 1095 | + } | |
| 1096 | + } | |
| 1097 | + return false; | |
| 1098 | + } | |
| 1099 | + | |
| 1100 | + /** | |
| 1101 | + * Score for an image whose size we can't read at all. | |
| 1102 | + * | |
| 1103 | + * Deliberately non-zero: an unmeasurable image must still beat nothing and | |
| 1104 | + * still be preloadable on a page where no image declares its size. But it | |
| 1105 | + * sits below a 200×200 declared area (40 000), so anything we CAN measure | |
| 1106 | + * as a plausible hero outranks a guess. | |
| 1107 | + */ | |
| 1108 | + private const UNKNOWN_SIZE_SCORE = 1; | |
| 1109 | + | |
| 1110 | + /** | |
| 1111 | + * Height-to-width ratio assumed when only a `w` descriptor is readable. | |
| 1112 | + * | |
| 1113 | + * 9/16 — the commonest hero/banner shape, and close enough that a | |
| 1114 | + * width-only candidate is compared against a declared w×h area on the | |
| 1115 | + * same scale rather than being systematically inflated. | |
| 1116 | + */ | |
| 1117 | + private const ASSUMED_ASPECT_RATIO = 9 / 16; | |
| 1118 | + | |
| 1119 | + /** | |
| 1120 | + * Largest `w` descriptor in a srcset, or 0 when there isn't one. | |
| 1121 | + * | |
| 1122 | + * Only `w` descriptors are read. An `x` descriptor (`hero.jpg 2x`) | |
| 1123 | + * describes pixel density, not layout width, so it says nothing about | |
| 1124 | + * rendered area. | |
| 1125 | + */ | |
| 1126 | + private static function widest_srcset_width( string $srcset ): int { | |
| 1127 | + if ( '' === $srcset ) { | |
| 1128 | + return 0; | |
| 1129 | + } | |
| 1130 | + $widest = 0; | |
| 1131 | + foreach ( explode( ',', $srcset ) as $candidate ) { | |
| 1132 | + if ( preg_match( '#(\d+)w\s*$#', trim( $candidate ), $m ) ) { | |
| 1133 | + $widest = max( $widest, (int) $m[1] ); | |
| 1134 | + } | |
| 1135 | + } | |
| 1136 | + return $widest; | |
| 1137 | + } | |
| 1138 | + | |
| 1139 | + /** | |
| 243 | 1140 | * Assemble one <link rel="preload" as="image" fetchpriority="high">. |
| 244 | 1141 | * |
| 245 | 1142 | * The href/srcset run through the `xspeed_lcp_preload_url` / |
| 246 | 1143 | * `xspeed_lcp_preload_srcset` filters first. This is the coordination point |
| @@ -251,9 +1148,9 @@ | ||
| 251 | 1148 | * resource goes un-preloaded). By filtering the URL, a webp/avif layer can |
| 252 | 1149 | * redirect the preload to the format it will actually serve, WITHOUT Free |
| 253 | 1150 | * knowing that layer exists. (FBS-83553 H3) |
| 254 | 1151 | */ |
| 255 | - private static function preload_link( string $src, string $srcset, string $sizes ): string { | |
| 1152 | + private static function preload_link( string $src, string $srcset, string $sizes, string $kind = 'img', string $media = '' ): string { | |
| 256 | 1153 | // Resolve the `type` from the ORIGINAL image URL (before rewriting), so a |
| 257 | 1154 | // negotiating layer can key off the source .jpg/.png — after rewriting, |
| 258 | 1155 | // the URL is already a .webp and the derivation would no-op. |
| 259 | 1156 | $original = $src; |
| @@ -264,20 +1161,28 @@ | ||
| 264 | 1161 | * even though the markup is baked into a shared cache file. |
| 265 | 1162 | * |
| 266 | 1163 | * @param string $type Defaults to '' (no type attribute). |
| 267 | 1164 | * @param string $src The ORIGINAL (pre-rewrite) preload URL. |
| 1165 | + * @param string $kind `img` for an <img>, `background` for a CSS | |
| 1166 | + * background or a video poster. Only an <img> can be | |
| 1167 | + * wrapped in <picture>; a background is requested by | |
| 1168 | + * the URL its stylesheet names. | |
| 268 | 1169 | */ |
| 269 | - $type = (string) apply_filters( 'xspeed_lcp_preload_type', '', $original ); | |
| 1170 | + $type = (string) apply_filters( 'xspeed_lcp_preload_type', '', $original, $kind ); | |
| 270 | 1171 | /** |
| 271 | 1172 | * Filter the LCP preload href. Return a modern-format sibling (webp/avif) |
| 272 | 1173 | * when one will actually be served for this image. |
| 273 | 1174 | * |
| 274 | - * @param string $src The original image URL chosen for preload. | |
| 1175 | + * @param string $src The original image URL chosen for preload. | |
| 1176 | + * @param string $kind `img` or `background`, as for xspeed_lcp_preload_type. | |
| 275 | 1177 | */ |
| 276 | - $src = (string) apply_filters( 'xspeed_lcp_preload_url', $src ); | |
| 1178 | + $src = (string) apply_filters( 'xspeed_lcp_preload_url', $src, $kind ); | |
| 277 | 1179 | if ( '' !== $srcset ) { |
| 278 | - /** @param string $srcset The original srcset chosen for preload. */ | |
| 279 | - $srcset = (string) apply_filters( 'xspeed_lcp_preload_srcset', $srcset ); | |
| 1180 | + /** | |
| 1181 | + * @param string $srcset The original srcset chosen for preload. | |
| 1182 | + * @param string $kind `img` or `background`. | |
| 1183 | + */ | |
| 1184 | + $srcset = (string) apply_filters( 'xspeed_lcp_preload_srcset', $srcset, $kind ); | |
| 280 | 1185 | } |
| 281 | 1186 | |
| 282 | 1187 | $attrs = sprintf( 'href="%s"', esc_url( $src ) ); |
| 283 | 1188 | |
| @@ -293,8 +1198,12 @@ | ||
| 293 | 1198 | if ( '' !== $type ) { |
| 294 | 1199 | $attrs .= sprintf( ' type="%s"', esc_attr( $type ) ); |
| 295 | 1200 | } |
| 296 | 1201 | |
| 1202 | + if ( '' !== $media ) { | |
| 1203 | + $attrs .= sprintf( ' media="%s"', esc_attr( $media ) ); | |
| 1204 | + } | |
| 1205 | + | |
| 297 | 1206 | return sprintf( '<link rel="preload" as="image" %s fetchpriority="high">' . "\n", $attrs ); |
| 298 | 1207 | } |
| 299 | 1208 | |
| 300 | 1209 | /** |
| @@ -301,9 +1210,17 @@ | ||
| 301 | 1210 | * Extract a single/double-quoted attribute value from a tag. Returns '' |
| 302 | 1211 | * when the attribute is absent. |
| 303 | 1212 | */ |
| 304 | 1213 | private static function attr( string $tag, string $name ): string { |
| 305 | - if ( preg_match( '#\b' . preg_quote( $name, '#' ) . '=(["\'])(.*?)\1#is', $tag, $m ) ) { | |
| 1214 | + // Anchor on a real attribute boundary, not `\b`. A word boundary sits | |
| 1215 | + // between the `-` and the `w` of `data-width`, so `\bwidth=` matched | |
| 1216 | + // inside it: a lazy-loaded hero carrying `data-width="50" | |
| 1217 | + // data-height="50"` was scored 50×50 and rejected by | |
| 1218 | + // looks_too_small() — defeating the feature on exactly the images the | |
| 1219 | + // data-src/data-srcset handling exists to support. Requiring | |
| 1220 | + // whitespace (or the start of the string) before the name means only | |
| 1221 | + // a genuine attribute matches. | |
| 1222 | + if ( preg_match( '#(?:^|\s)' . preg_quote( $name, '#' ) . '\s*=\s*(["\'])(.*?)\1#is', $tag, $m ) ) { | |
| 306 | 1223 | return trim( $m[2] ); |
| 307 | 1224 | } |
| 308 | 1225 | return ''; |
| 309 | 1226 | } |
| @@ -354,13 +1271,21 @@ | ||
| 354 | 1271 | */ |
| 355 | 1272 | private const NON_HERO_MARKERS = array( 'logo', 'icon', 'avatar', 'gravatar', 'spinner', 'emoji', 'site-icon', 'custom-logo' ); |
| 356 | 1273 | |
| 357 | 1274 | /** |
| 1275 | + * Below this declared area (px²) an image is a badge/thumb/divider, never | |
| 1276 | + * an LCP hero — 10 000 is a 100×100 square, or a 500×20 strip. Applied | |
| 1277 | + * only when BOTH dimensions are readable. (FBS-84576) | |
| 1278 | + */ | |
| 1279 | + private const MIN_LCP_AREA = 10000; | |
| 1280 | + | |
| 1281 | + /** | |
| 358 | 1282 | * Is this <img> too small / too chrome-like to be the LCP hero? True when |
| 359 | 1283 | * either (a) it carries a logo/icon/avatar marker, (b) an explicit |
| 360 | - * `data-no-lcp` opt-out, or (c) BOTH width and height are present and both | |
| 361 | - * are ≤ the threshold. Missing dimensions are NOT guessed — an image whose | |
| 362 | - * size we can't read still competes. (FBS-83553 H1 "logo before hero".) | |
| 1284 | + * `data-no-lcp` opt-out, or (c) BOTH width and height are present and | |
| 1285 | + * both are ≤ the dimension threshold, or their area is under | |
| 1286 | + * MIN_LCP_AREA. Missing dimensions are NOT guessed — an image whose size | |
| 1287 | + * we can't read still competes. (FBS-83553 H1 "logo before hero".) | |
| 363 | 1288 | */ |
| 364 | 1289 | private static function looks_too_small( string $tag ): bool { |
| 365 | 1290 | if ( false !== stripos( $tag, 'data-no-lcp' ) ) { |
| 366 | 1291 | return true; |
| @@ -377,12 +1302,232 @@ | ||
| 377 | 1302 | $h = self::attr( $tag, 'height' ); |
| 378 | 1303 | if ( '' === $w || '' === $h || ! is_numeric( $w ) || ! is_numeric( $h ) ) { |
| 379 | 1304 | return false; // unknown size — don't guess; let it compete. |
| 380 | 1305 | } |
| 1306 | + if ( (int) $w * (int) $h < self::MIN_LCP_AREA ) { | |
| 1307 | + return true; | |
| 1308 | + } | |
| 381 | 1309 | return (int) $w <= self::MIN_LCP_DIMENSION && (int) $h <= self::MIN_LCP_DIMENSION; |
| 382 | 1310 | } |
| 383 | 1311 | |
| 1312 | + /** Most candidates one answer may carry (one per device class). */ | |
| 1313 | + private const MAX_SUPPLIED_CANDIDATES = 3; | |
| 1314 | + | |
| 384 | 1315 | /** |
| 1316 | + * The filter's answer as a list of well-formed candidates, or null when | |
| 1317 | + * it is no answer or not one this understands (then the automatic pick | |
| 1318 | + * runs). An empty list means "this page has no LCP image". | |
| 1319 | + * | |
| 1320 | + * @param mixed $answer What `xspeed_lcp_preload_candidate` returned. | |
| 1321 | + * @return array<int,array{url:string,srcset:string,sizes:string,kind:string,media:string}>|null | |
| 1322 | + */ | |
| 1323 | + private static function supplied_candidates( $answer ): ?array { | |
| 1324 | + if ( ! is_array( $answer ) || empty( $answer ) ) { | |
| 1325 | + return null; | |
| 1326 | + } | |
| 1327 | + $items = array_key_exists( 'url', $answer ) ? array( $answer ) : $answer; | |
| 1328 | + if ( array_values( $items ) !== $items || count( $items ) > self::MAX_SUPPLIED_CANDIDATES ) { | |
| 1329 | + return null; | |
| 1330 | + } | |
| 1331 | + $out = array(); | |
| 1332 | + foreach ( $items as $item ) { | |
| 1333 | + if ( ! is_array( $item ) || ! array_key_exists( 'url', $item ) || ! is_string( $item['url'] ) ) { | |
| 1334 | + return null; | |
| 1335 | + } | |
| 1336 | + $url = trim( $item['url'] ); | |
| 1337 | + if ( '' === $url ) { | |
| 1338 | + // "No LCP image" only makes sense as the whole answer. | |
| 1339 | + if ( 1 === count( $items ) ) { | |
| 1340 | + return array(); | |
| 1341 | + } | |
| 1342 | + return null; | |
| 1343 | + } | |
| 1344 | + // Absolute, protocol-relative or root-relative. A bare `hero.png` | |
| 1345 | + // passed esc_url() as `http://hero.png`: a high-priority fetch to | |
| 1346 | + // a host that does not exist. | |
| 1347 | + if ( '' === esc_url( $url ) || ! preg_match( '#^(?:https?:)?/#i', $url ) ) { | |
| 1348 | + return null; | |
| 1349 | + } | |
| 1350 | + // A media rule this cannot use rejects the answer. Dropping only | |
| 1351 | + // the rule kept the preload and sent it to every device. | |
| 1352 | + $media = $item['media'] ?? ''; | |
| 1353 | + if ( ! is_string( $media ) || strlen( $media ) > 200 || preg_match( '#[<>"]#', $media ) ) { | |
| 1354 | + return null; | |
| 1355 | + } | |
| 1356 | + $text = static fn ( $v, int $max ): string => ( is_string( $v ) && strlen( $v ) <= $max ) ? trim( $v ) : ''; | |
| 1357 | + $out[] = array( | |
| 1358 | + 'url' => $url, | |
| 1359 | + 'srcset' => $text( $item['srcset'] ?? '', 4096 ), | |
| 1360 | + 'sizes' => $text( $item['sizes'] ?? '', 512 ), | |
| 1361 | + 'kind' => 'background' === ( $item['kind'] ?? '' ) ? 'background' : 'img', | |
| 1362 | + 'media' => trim( $media ), | |
| 1363 | + ); | |
| 1364 | + } | |
| 1365 | + return $out; | |
| 1366 | + } | |
| 1367 | + | |
| 1368 | + /** | |
| 1369 | + * Preload what an extension supplied, and promote the <img> it names. | |
| 1370 | + * | |
| 1371 | + * The answer is measured, not guessed, so every other <img> gives up a | |
| 1372 | + * stray `fetchpriority="high"`: two High images compete with the one | |
| 1373 | + * that paints first. | |
| 1374 | + * | |
| 1375 | + * @param string $html Page HTML. | |
| 1376 | + * @param array<int,array{url:string,srcset:string,sizes:string,kind:string,media:string}> $candidates From supplied_candidates(). | |
| 1377 | + * @return array{0:string,1:string} [rewritten html, preload markup] | |
| 1378 | + */ | |
| 1379 | + private static function build_supplied_preload( string $html, array $candidates ): array { | |
| 1380 | + // Scoped when any preload carries a media rule, even a single one: | |
| 1381 | + // Pro sends one desktop-only candidate when the phone's LCP is text, | |
| 1382 | + // and promoting that <img> made phones download a hero they hide. | |
| 1383 | + $scoped = '' !== implode( '', array_column( $candidates, 'media' ) ); | |
| 1384 | + $skip = array_merge( self::chrome_container_ranges( $html ), Lazy_Loader::hidden_ranges( $html ) ); | |
| 1385 | + | |
| 1386 | + // Which <img> index each candidate promotes: the first one, outside | |
| 1387 | + // site chrome and hidden markup, whose src or one of whose srcset | |
| 1388 | + // URLs IS the candidate. A substring match picked an earlier image | |
| 1389 | + // whose srcset merely contained the name. | |
| 1390 | + // Two candidates may claim the same <img>: a phone and a desktop size | |
| 1391 | + // of one responsive image. The second one is still an <img>, not a | |
| 1392 | + // background. | |
| 1393 | + $promote = array(); | |
| 1394 | + $matched = array(); | |
| 1395 | + if ( preg_match_all( '#<img\b[^>]*>#i', $html, $imgs, PREG_OFFSET_CAPTURE ) ) { | |
| 1396 | + foreach ( $candidates as $c => $candidate ) { | |
| 1397 | + foreach ( $imgs[0] as $index => $match ) { | |
| 1398 | + [ $tag, $offset ] = $match; | |
| 1399 | + if ( self::offset_in_ranges( $offset, $skip ) || Lazy_Loader::tag_is_hidden( $tag, 'img' ) ) { | |
| 1400 | + continue; | |
| 1401 | + } | |
| 1402 | + [ $src, $img_srcset, $img_sizes ] = self::effective_image_src( $tag ); | |
| 1403 | + if ( ! self::same_url( $src, $candidate['url'] ) && ! self::srcset_has( $img_srcset, $candidate['url'] ) ) { | |
| 1404 | + continue; | |
| 1405 | + } | |
| 1406 | + $promote[ $index ] = true; | |
| 1407 | + $matched[ $c ] = true; | |
| 1408 | + // The <img> may already be served as a format sibling the | |
| 1409 | + // answer did not name (`hero.png.webp` for `hero.png`: an | |
| 1410 | + // image plugin swapped it for this browser). Preload what | |
| 1411 | + // the <img> will fetch, or the page downloads both. | |
| 1412 | + if ( self::format_original( $src ) !== $src ) { | |
| 1413 | + $candidates[ $c ]['url'] = html_entity_decode( $src, ENT_QUOTES ); | |
| 1414 | + $candidates[ $c ]['srcset'] = html_entity_decode( $img_srcset, ENT_QUOTES ); | |
| 1415 | + $candidates[ $c ]['sizes'] = $img_sizes; | |
| 1416 | + } | |
| 1417 | + // An answer that names only the URL still gets the <img>'s | |
| 1418 | + // responsive set: without it the preload fetches the full | |
| 1419 | + // file and the browser then downloads the size it shows. | |
| 1420 | + if ( '' === $candidates[ $c ]['srcset'] && '' !== $img_srcset ) { | |
| 1421 | + $candidates[ $c ]['srcset'] = $img_srcset; | |
| 1422 | + $candidates[ $c ]['sizes'] = $img_sizes; | |
| 1423 | + } | |
| 1424 | + break; | |
| 1425 | + } | |
| 1426 | + } | |
| 1427 | + } | |
| 1428 | + | |
| 1429 | + $seen = -1; | |
| 1430 | + $html = (string) preg_replace_callback( | |
| 1431 | + '#<img\b[^>]*>#i', | |
| 1432 | + static function ( array $m ) use ( &$seen, $promote, $scoped ) { | |
| 1433 | + ++$seen; | |
| 1434 | + if ( isset( $promote[ $seen ] ) ) { | |
| 1435 | + // Scoped to a device: make the <img> lazy, so the device | |
| 1436 | + // that hides it never downloads it. Keeping whatever | |
| 1437 | + // `loading` it had was not enough: Lazy Load's default | |
| 1438 | + // "load the first image straight away" had already made | |
| 1439 | + // it eager. The device that shows it gets it from its own | |
| 1440 | + // preload, which already fetched it early. | |
| 1441 | + return $scoped ? self::force_lazy( self::set_fetchpriority( $m[0] ) ) : self::promote_lcp_img( $m[0] ); | |
| 1442 | + } | |
| 1443 | + if ( 'high' === strtolower( self::attr( $m[0], 'fetchpriority' ) ) ) { | |
| 1444 | + return (string) preg_replace( '#\s*(?<![-\w])fetchpriority\s*=\s*(["\']?)high\1#i', '', $m[0], 1 ); | |
| 1445 | + } | |
| 1446 | + return $m[0]; | |
| 1447 | + }, | |
| 1448 | + $html | |
| 1449 | + ); | |
| 1450 | + | |
| 1451 | + $preload = ''; | |
| 1452 | + foreach ( $candidates as $c => $candidate ) { | |
| 1453 | + // Idempotency, as for the automatic pick: a second pass over an | |
| 1454 | + // already-processed body must not emit the link twice. Compared | |
| 1455 | + // with `&` decoded, the way the markup spells a query. | |
| 1456 | + $already = false; | |
| 1457 | + if ( preg_match_all( '#<link\b[^>]*rel=["\']preload["\'][^>]*>#i', $html, $links ) ) { | |
| 1458 | + foreach ( $links[0] as $link ) { | |
| 1459 | + if ( false !== strpos( html_entity_decode( $link, ENT_QUOTES ), $candidate['url'] ) ) { | |
| 1460 | + $already = true; | |
| 1461 | + break; | |
| 1462 | + } | |
| 1463 | + } | |
| 1464 | + } | |
| 1465 | + if ( $already ) { | |
| 1466 | + continue; | |
| 1467 | + } | |
| 1468 | + // An image the page shows through no <img> is a background, | |
| 1469 | + // whatever the answer said: nothing here can wrap it in <picture>. | |
| 1470 | + $kind = isset( $matched[ $c ] ) ? $candidate['kind'] : 'background'; | |
| 1471 | + $preload .= self::preload_link( $candidate['url'], $candidate['srcset'], $candidate['sizes'], $kind, $candidate['media'] ); | |
| 1472 | + } | |
| 1473 | + return array( $html, $preload ); | |
| 1474 | + } | |
| 1475 | + | |
| 1476 | + /** Set `loading="lazy"` on an <img>, replacing any other loading value. */ | |
| 1477 | + private static function force_lazy( string $tag ): string { | |
| 1478 | + $tag = (string) preg_replace( '#\s*(?<![-\w])loading\s*=\s*(["\']?)[a-z]*\1#i', '', $tag, 1 ); | |
| 1479 | + return (string) preg_replace( '#<img\b#i', '<img loading="lazy"', $tag, 1 ); | |
| 1480 | + } | |
| 1481 | + | |
| 1482 | + /** Two URLs name the same image: `&` decoded, and a root-relative one compared by path. */ | |
| 1483 | + private static function same_url( string $a, string $b ): bool { | |
| 1484 | + $a = html_entity_decode( $a, ENT_QUOTES ); | |
| 1485 | + $b = html_entity_decode( $b, ENT_QUOTES ); | |
| 1486 | + if ( '' === $a || '' === $b ) { | |
| 1487 | + return false; | |
| 1488 | + } | |
| 1489 | + if ( $a === $b ) { | |
| 1490 | + return true; | |
| 1491 | + } | |
| 1492 | + // A format sibling is the same image: `hero.png.webp` is `hero.png`. | |
| 1493 | + $a = self::format_original( $a ); | |
| 1494 | + $b = self::format_original( $b ); | |
| 1495 | + if ( $a === $b ) { | |
| 1496 | + return true; | |
| 1497 | + } | |
| 1498 | + $rel = static function ( string $u ): string { | |
| 1499 | + if ( 0 === strpos( $u, '/' ) && 0 !== strpos( $u, '//' ) ) { | |
| 1500 | + return $u; | |
| 1501 | + } | |
| 1502 | + $path = (string) wp_parse_url( $u, PHP_URL_PATH ); | |
| 1503 | + $query = (string) wp_parse_url( $u, PHP_URL_QUERY ); | |
| 1504 | + return $path . ( '' !== $query ? '?' . $query : '' ); | |
| 1505 | + }; | |
| 1506 | + return ( 0 === strpos( $a, '/' ) || 0 === strpos( $b, '/' ) ) && $rel( $a ) === $rel( $b ); | |
| 1507 | + } | |
| 1508 | + | |
| 1509 | + /** | |
| 1510 | + * The original of a modern-format sibling that appends its extension, | |
| 1511 | + * `hero.png.webp` → `hero.png`, the naming ShortPixel, Imagify and other | |
| 1512 | + * image plugins use. Any other URL comes back unchanged. | |
| 1513 | + */ | |
| 1514 | + private static function format_original( string $url ): string { | |
| 1515 | + return (string) preg_replace( '#(\.(?:jpe?g|png))\.(?:webp|avif)(?=$|\?)#i', '$1', $url ); | |
| 1516 | + } | |
| 1517 | + | |
| 1518 | + /** Whether one of the URLs in a srcset IS $url (not merely contains it). */ | |
| 1519 | + private static function srcset_has( string $srcset, string $url ): bool { | |
| 1520 | + foreach ( explode( ',', $srcset ) as $entry ) { | |
| 1521 | + $candidate = strtok( trim( $entry ), " \t\n" ); | |
| 1522 | + if ( false !== $candidate && self::same_url( $candidate, $url ) ) { | |
| 1523 | + return true; | |
| 1524 | + } | |
| 1525 | + } | |
| 1526 | + return false; | |
| 1527 | + } | |
| 1528 | + | |
| 1529 | + /** | |
| 385 | 1530 | * Promote an <img> to the LCP element: force fetchpriority="high" and |
| 386 | 1531 | * strip any loading="lazy" so the browser loads it immediately. Both are |
| 387 | 1532 | * idempotent. `loading="lazy"` is REMOVED rather than flipped to "eager" |
| 388 | 1533 | * because eager is the default; a bare tag with fetchpriority="high" is |
| @@ -413,8 +1558,16 @@ | ||
| 413 | 1558 | * before the first stylesheet so the preloads are discovered before the |
| 414 | 1559 | * render-blocking CSS. Falls back to after <head>, then prepend. |
| 415 | 1560 | */ |
| 416 | 1561 | private static function inject_into_head( string $html, string $hints ): string { |
| 1562 | + // Only a supplied candidate's preload carries `media`. Every other | |
| 1563 | + // page keeps exactly the placement it had before the seam existed. | |
| 1564 | + if ( false !== strpos( $hints, ' media="' ) ) { | |
| 1565 | + $pos = self::after_viewport_offset( $html ); | |
| 1566 | + if ( null !== $pos ) { | |
| 1567 | + return substr( $html, 0, $pos ) . "\n" . $hints . substr( $html, $pos ); | |
| 1568 | + } | |
| 1569 | + } | |
| 417 | 1570 | // Before the first <link rel="stylesheet"> if there is one. |
| 418 | 1571 | if ( preg_match( '#<link\b[^>]*rel=["\']stylesheet["\'][^>]*>#i', $html, $m, PREG_OFFSET_CAPTURE ) ) { |
| 419 | 1572 | $pos = $m[0][1]; |
| 420 | 1573 | return substr( $html, 0, $pos ) . $hints . substr( $html, $pos ); |
| @@ -425,6 +1578,147 @@ | ||
| 425 | 1578 | return substr( $html, 0, $pos ) . "\n" . $hints . substr( $html, $pos ); |
| 426 | 1579 | } |
| 427 | 1580 | // No head at all — prepend (degenerate documents). |
| 428 | 1581 | return $hints . $html; |
| 1582 | + } | |
| 1583 | + | |
| 1584 | + /** | |
| 1585 | + * Where a media-scoped preload goes: before the first stylesheet, as | |
| 1586 | + * usual, but never ahead of <meta name="viewport">. Null when the head | |
| 1587 | + * could not be read, and the usual placement is used instead. | |
| 1588 | + * | |
| 1589 | + * A preload's `media` is evaluated when the parser reaches the link, and | |
| 1590 | + * until the viewport meta is parsed a phone lays out at its default | |
| 1591 | + * desktop width. A `(min-width: 768px)` preload placed earlier matched on | |
| 1592 | + * phones too and fetched the desktop hero at high priority. | |
| 1593 | + * | |
| 1594 | + * Reads only the head, tag by tag, so a viewport meta or a stylesheet | |
| 1595 | + * written inside a script string, a comment or an attribute value is | |
| 1596 | + * never taken for a real one. That put the hints inside a script, which | |
| 1597 | + * then threw a SyntaxError. A pattern over the whole page did the same | |
| 1598 | + * job but could give up on a large inline block, and then searched the | |
| 1599 | + * raw markup. | |
| 1600 | + * | |
| 1601 | + * @return int|null Byte offset in $html. | |
| 1602 | + */ | |
| 1603 | + private static function after_viewport_offset( string $html ) { | |
| 1604 | + $len = strlen( $html ); | |
| 1605 | + $i = 0; | |
| 1606 | + $head_start = null; | |
| 1607 | + $stylesheet = null; | |
| 1608 | + $viewport = null; | |
| 1609 | + while ( $i < $len ) { | |
| 1610 | + $lt = strpos( $html, '<', $i ); | |
| 1611 | + if ( false === $lt ) { | |
| 1612 | + break; | |
| 1613 | + } | |
| 1614 | + if ( 0 === substr_compare( $html, '<!--', $lt, 4 ) ) { | |
| 1615 | + // `<!-->` and `<!--->` close at once, so look from the second dash. | |
| 1616 | + $close = strpos( $html, '-->', $lt + 2 ); | |
| 1617 | + if ( false === $close ) { | |
| 1618 | + return null; | |
| 1619 | + } | |
| 1620 | + $i = $close + 3; | |
| 1621 | + continue; | |
| 1622 | + } | |
| 1623 | + if ( ! preg_match( '#\G<(/?)([a-z][a-z0-9-]*)#i', $html, $t, 0, $lt ) ) { | |
| 1624 | + $i = $lt + 1; | |
| 1625 | + continue; | |
| 1626 | + } | |
| 1627 | + $gt = self::tag_end( $html, $lt + strlen( $t[0] ) ); | |
| 1628 | + if ( null === $gt ) { | |
| 1629 | + return null; | |
| 1630 | + } | |
| 1631 | + $name = strtolower( $t[2] ); | |
| 1632 | + $tag = substr( $html, $lt, $gt + 1 - $lt ); | |
| 1633 | + $i = $gt + 1; | |
| 1634 | + | |
| 1635 | + if ( '/' === $t[1] ) { | |
| 1636 | + if ( 'head' === $name ) { | |
| 1637 | + break; | |
| 1638 | + } | |
| 1639 | + continue; | |
| 1640 | + } | |
| 1641 | + if ( 'head' === $name ) { | |
| 1642 | + $head_start = $i; | |
| 1643 | + continue; | |
| 1644 | + } | |
| 1645 | + if ( 'body' === $name ) { | |
| 1646 | + break; | |
| 1647 | + } | |
| 1648 | + if ( in_array( $name, array( 'script', 'style', 'noscript', 'template', 'title', 'textarea' ), true ) ) { | |
| 1649 | + $close = stripos( $html, '</' . $name, $i ); | |
| 1650 | + if ( false === $close ) { | |
| 1651 | + return null; | |
| 1652 | + } | |
| 1653 | + $i = $close; | |
| 1654 | + continue; | |
| 1655 | + } | |
| 1656 | + if ( null === $head_start ) { | |
| 1657 | + continue; | |
| 1658 | + } | |
| 1659 | + if ( 'link' !== $name && 'meta' !== $name ) { | |
| 1660 | + continue; | |
| 1661 | + } | |
| 1662 | + $attrs = self::tag_attrs( $tag ); | |
| 1663 | + if ( null === $stylesheet && 'link' === $name && 'stylesheet' === strtolower( $attrs['rel'] ?? '' ) ) { | |
| 1664 | + if ( null !== $viewport ) { | |
| 1665 | + return $lt; | |
| 1666 | + } | |
| 1667 | + $stylesheet = $lt; | |
| 1668 | + continue; | |
| 1669 | + } | |
| 1670 | + if ( null === $viewport && 'meta' === $name && 'viewport' === strtolower( $attrs['name'] ?? '' ) ) { | |
| 1671 | + if ( null !== $stylesheet ) { | |
| 1672 | + return $i; | |
| 1673 | + } | |
| 1674 | + $viewport = $i; | |
| 1675 | + } | |
| 1676 | + } | |
| 1677 | + if ( null !== $stylesheet ) { | |
| 1678 | + return $stylesheet; | |
| 1679 | + } | |
| 1680 | + return $viewport ?? $head_start; | |
| 1681 | + } | |
| 1682 | + | |
| 1683 | + /** | |
| 1684 | + * A tag's attributes, read in order so a quoted value is consumed whole: | |
| 1685 | + * `content="name=viewport"` is a content attribute, not a name. | |
| 1686 | + * | |
| 1687 | + * @return array<string,string> Lowercase name => trimmed value; the first of a repeated name wins. | |
| 1688 | + */ | |
| 1689 | + private static function tag_attrs( string $tag ): array { | |
| 1690 | + $out = array(); | |
| 1691 | + preg_match_all( '#\s([^\s=/>"\']+)(?:\s*=\s*("[^"]*"|\'[^\']*\'|[^\s>]+))?#', $tag, $m, PREG_SET_ORDER ); | |
| 1692 | + foreach ( $m as $a ) { | |
| 1693 | + $key = strtolower( $a[1] ); | |
| 1694 | + if ( ! isset( $out[ $key ] ) ) { | |
| 1695 | + $out[ $key ] = trim( trim( $a[2] ?? '', '"\'' ) ); | |
| 1696 | + } | |
| 1697 | + } | |
| 1698 | + return $out; | |
| 1699 | + } | |
| 1700 | + | |
| 1701 | + /** | |
| 1702 | + * Offset of the `>` that ends a tag whose attributes start at $from, | |
| 1703 | + * stepping over quoted values so `content="a>b"` does not end it early. | |
| 1704 | + * | |
| 1705 | + * @return int|null | |
| 1706 | + */ | |
| 1707 | + private static function tag_end( string $html, int $from ) { | |
| 1708 | + $len = strlen( $html ); | |
| 1709 | + $quote = ''; | |
| 1710 | + for ( $j = $from; $j < $len; $j++ ) { | |
| 1711 | + $c = $html[ $j ]; | |
| 1712 | + if ( '' !== $quote ) { | |
| 1713 | + if ( $c === $quote ) { | |
| 1714 | + $quote = ''; | |
| 1715 | + } | |
| 1716 | + } elseif ( '"' === $c || "'" === $c ) { | |
| 1717 | + $quote = $c; | |
| 1718 | + } elseif ( '>' === $c ) { | |
| 1719 | + return $j; | |
| 1720 | + } | |
| 1721 | + } | |
| 1722 | + return null; | |
| 429 | 1723 | } |
| 430 | 1724 | } |
