| @@ -52,11 +52,86 @@ | ||
| 52 | 52 | '_breakdance_data', // Oxygen 6+ / Breakdance |
| 53 | 53 | '_oxygen_data', // Oxygen (earlier releases) |
| 54 | 54 | 'ct_builder_shortcodes', // Oxygen classic |
| 55 | 55 | '_elementor_data', // Elementor |
| 56 | + // Beaver Builder. Published layout first: `_fl_builder_draft` holds | |
| 57 | + // unsaved changes and would score content the visitor cannot see. | |
| 58 | + // Both are arrays of stdClass nodes, which is why the walker below | |
| 59 | + // has to treat objects like arrays (#449). | |
| 60 | + '_fl_builder_data', // Beaver Builder (published) | |
| 61 | + '_fl_builder_draft', // Beaver Builder (unsaved changes) | |
| 56 | 62 | ]; |
| 57 | 63 | |
| 58 | 64 | /** |
| 65 | + * Bricks' content-area meta key, used when Bricks itself isn't loaded. | |
| 66 | + * | |
| 67 | + * Bricks exposes `BRICKS_DB_PAGE_CONTENT` and renames the underlying key | |
| 68 | + * between generations (it gained the `_2` suffix in 1.7.3), so the | |
| 69 | + * constant is authoritative and this literal is only the fallback for the | |
| 70 | + * contexts where it is undefined — Bricks is a theme, so on an admin or | |
| 71 | + * CLI request against a site that has since switched themes the constant | |
| 72 | + * is simply not there while the post meta still is. | |
| 73 | + * | |
| 74 | + * Bricks stores three areas — header, content and footer. Only the content | |
| 75 | + * area belongs to the post being scored; the header and footer areas live | |
| 76 | + * on Bricks' own template posts and would double-count site chrome into | |
| 77 | + * every page's word count, so they are deliberately not read here. | |
| 78 | + * | |
| 79 | + * @since 2.2.1 | |
| 80 | + * @var string | |
| 81 | + */ | |
| 82 | + private const BRICKS_CONTENT_META_KEY = '_bricks_page_content_2'; | |
| 83 | + | |
| 84 | + /** | |
| 85 | + * Bricks' per-post editor-mode meta key, used when Bricks isn't loaded. | |
| 86 | + * | |
| 87 | + * @since 2.2.1 | |
| 88 | + * @var string | |
| 89 | + */ | |
| 90 | + private const BRICKS_EDITOR_MODE_META_KEY = '_bricks_editor_mode'; | |
| 91 | + | |
| 92 | + /** | |
| 93 | + * Bricks' components option, used when Bricks itself isn't loaded. | |
| 94 | + * | |
| 95 | + * @since 2.2.1 | |
| 96 | + * @var string | |
| 97 | + */ | |
| 98 | + private const BRICKS_COMPONENTS_OPTION = 'bricks_components'; | |
| 99 | + | |
| 100 | + /** | |
| 101 | + * Bricks' element that renders the post's own `post_content`. | |
| 102 | + * | |
| 103 | + * A Bricks page normally discards `post_content` entirely, which is why | |
| 104 | + * anything left there is invisible. Dropping this element onto the canvas | |
| 105 | + * is the one way an author puts it back on the page, so its presence flips | |
| 106 | + * `post_content` from stale leftovers to content the visitor reads. | |
| 107 | + * | |
| 108 | + * @since 2.3.1 | |
| 109 | + * @var string | |
| 110 | + */ | |
| 111 | + private const BRICKS_POST_CONTENT_ELEMENT = 'post-content'; | |
| 112 | + | |
| 113 | + /** | |
| 114 | + * Resolved Bricks trees for this request, keyed by post ID. | |
| 115 | + * | |
| 116 | + * Rendering one page asks for the tree about twenty times — every | |
| 117 | + * description, every schema node, the FAQ guard — and resolving it is not | |
| 118 | + * free. `bricks_content_source()` clears `Bricks\Database::$active_templates` | |
| 119 | + * before asking Bricks which content template applies, which defeats | |
| 120 | + * Bricks' own early-return and re-runs its whole template-condition engine; | |
| 121 | + * `expand_bricks_components()` then walks the tree again. Measured on a | |
| 122 | + * Bricks page with no content of its own, that was ten full runs of the | |
| 123 | + * rules engine per request. | |
| 124 | + * | |
| 125 | + * Per-request only, and only ever read back within one page render — a | |
| 126 | + * request that writes Bricks content does not also render it. | |
| 127 | + * | |
| 128 | + * @since 2.3.1 | |
| 129 | + * @var array<int,array<int,mixed>> | |
| 130 | + */ | |
| 131 | + private static array $bricks_trees = []; | |
| 132 | + | |
| 133 | + /** | |
| 59 | 134 | * JSON keys whose values are user-visible text. |
| 60 | 135 | * |
| 61 | 136 | * Builder trees mix content with configuration, so a blind string sweep |
| 62 | 137 | * would count CSS classes and option slugs as words. Matching on the key |
| @@ -84,8 +159,71 @@ | ||
| 84 | 159 | 'link', 'url', 'href', 'link_url', 'button_link', 'permalink', 'link_to', |
| 85 | 160 | ]; |
| 86 | 161 | |
| 87 | 162 | /** |
| 163 | + * JSON keys whose values hold an embedded video's source. | |
| 164 | + * | |
| 165 | + * A builder's video widget keeps its destination in a provider-specific | |
| 166 | + * field — Elementor picks `youtube_url`, `vimeo_url`, `dailymotion_url` or | |
| 167 | + * `hosted_url` according to the chosen source type — none of which is a | |
| 168 | + * link field or a content field, so a video on a builder page reached the | |
| 169 | + * analyzers as nothing at all. | |
| 170 | + * | |
| 171 | + * These are deliberately kept out of URL_KEYS. A video is an embed, not an | |
| 172 | + * outbound link: rendering one as `<a href>` would add a spurious external | |
| 173 | + * link to every page carrying a video and skew the link counts. They are | |
| 174 | + * reconstructed as `<iframe>`/`<video>` instead, which the video detector | |
| 175 | + * recognises and the link and image counters ignore. | |
| 176 | + * | |
| 177 | + * @since 2.3.1 | |
| 178 | + * @var string[] | |
| 179 | + */ | |
| 180 | + private const VIDEO_KEYS = [ | |
| 181 | + 'youtube_url', 'vimeo_url', 'dailymotion_url', 'videopress_url', | |
| 182 | + 'hosted_url', 'video_url', 'video_src', 'video_link', | |
| 183 | + ]; | |
| 184 | + | |
| 185 | + /** | |
| 186 | + * File extensions that mean a video source is a file, not a provider page. | |
| 187 | + * | |
| 188 | + * @since 2.3.1 | |
| 189 | + * @var string[] | |
| 190 | + */ | |
| 191 | + private const VIDEO_FILE_EXTENSIONS = ['mp4', 'webm', 'ogv', 'mov', 'm4v']; | |
| 192 | + | |
| 193 | + /** | |
| 194 | + * Source keys to trust for a declared video source type. | |
| 195 | + * | |
| 196 | + * A widget keeps one field per provider and does not clear the others when | |
| 197 | + * the author switches source: an Elementor video moved from YouTube to Self | |
| 198 | + * Hosted still carries the earlier `youtube_url`. Reading whichever key | |
| 199 | + * turns up first then emits the video the author replaced. The widget says | |
| 200 | + * which one it is actually playing, so that is read first and the flat key | |
| 201 | + * sweep is only the fallback for a builder that declares nothing. | |
| 202 | + * | |
| 203 | + * @since 2.3.1 | |
| 204 | + * @var array<string,string[]> | |
| 205 | + */ | |
| 206 | + private const VIDEO_KEYS_BY_TYPE = [ | |
| 207 | + 'youtube' => ['youtube_url'], | |
| 208 | + 'vimeo' => ['vimeo_url'], | |
| 209 | + 'dailymotion' => ['dailymotion_url'], | |
| 210 | + 'videopress' => ['videopress_url'], | |
| 211 | + 'hosted' => ['hosted_url', 'video_url', 'video_src', 'video_link'], | |
| 212 | + 'media' => ['hosted_url', 'video_url', 'video_src', 'video_link'], | |
| 213 | + 'file' => ['hosted_url', 'video_url', 'video_src', 'video_link'], | |
| 214 | + 'self_hosted' => ['hosted_url', 'video_url', 'video_src', 'video_link'], | |
| 215 | + ]; | |
| 216 | + | |
| 217 | + /** | |
| 218 | + * Keys a builder uses to name which video source a widget is playing. | |
| 219 | + * | |
| 220 | + * @since 2.3.1 | |
| 221 | + * @var string[] | |
| 222 | + */ | |
| 223 | + private const VIDEO_TYPE_KEYS = ['video_type', 'videotype', 'video_source', 'source_type']; | |
| 224 | + | |
| 225 | + /** | |
| 88 | 226 | * JSON keys whose values hold an image, as a URL string or `{ url, alt }`. |
| 89 | 227 | * |
| 90 | 228 | * @var string[] |
| 91 | 229 | */ |
| @@ -102,9 +240,13 @@ | ||
| 102 | 240 | * |
| 103 | 241 | * @var string[] |
| 104 | 242 | */ |
| 105 | 243 | private const HEADING_TAG_KEYS = [ |
| 106 | - 'header_size', 'heading_tag', 'html_tag', 'title_tag', 'tag', 'level', 'size', | |
| 244 | + // Lower-cased on both sides of the comparison, so `headingtag` is the | |
| 245 | + // camelCase `headingTag` Bricks uses throughout its own controls and | |
| 246 | + // which ThinkRank's Bricks elements declare. Without it their section | |
| 247 | + // headings counted as body copy and never reached heading structure. | |
| 248 | + 'header_size', 'heading_tag', 'headingtag', 'html_tag', 'title_tag', 'tag', 'level', 'size', | |
| 107 | 249 | ]; |
| 108 | 250 | |
| 109 | 251 | /** |
| 110 | 252 | * Keys whose value is alternative text for a sibling image. |
| @@ -119,12 +261,231 @@ | ||
| 119 | 261 | * @param \WP_Post $post Post being analyzed. |
| 120 | 262 | * @return string HTML/text to analyze. |
| 121 | 263 | */ |
| 122 | 264 | public static function resolve(\WP_Post $post): string { |
| 123 | - return self::resolve_markup((string) $post->post_content, $post); | |
| 265 | + $raw = (string) $post->post_content; | |
| 266 | + | |
| 267 | + // A page built in Gutenberg and then switched to Bricks keeps its old | |
| 268 | + // blocks in `post_content` forever — Bricks never clears them, and | |
| 269 | + // never renders them either. Resolving that first meant the stale draft | |
| 270 | + // beat the tree the visitor actually reads, and it did not stop at the | |
| 271 | + // score: the same string becomes the meta description, og:description, | |
| 272 | + // twitter:description and the schema description. Starting from nothing | |
| 273 | + // sends the resolution straight to Bricks' storage, which is where this | |
| 274 | + // page's words are (#651). | |
| 275 | + // | |
| 276 | + // Only for the stored path. `resolve_markup()` is also called with live | |
| 277 | + // editor content, and the Bricks panel's resolver reads the canvas — | |
| 278 | + // discarding that would replace what the author is typing with the last | |
| 279 | + // save. | |
| 280 | + if (self::bricks_supersedes_post_content((int) $post->ID)) { | |
| 281 | + $raw = ''; | |
| 282 | + } | |
| 283 | + | |
| 284 | + return self::resolve_markup($raw, $post); | |
| 124 | 285 | } |
| 125 | 286 | |
| 126 | 287 | /** |
| 288 | + * Whether Bricks renders this post and throws its `post_content` away. | |
| 289 | + * | |
| 290 | + * True means anything still stored in `post_content` is invisible: it is | |
| 291 | + * not on the page, so it must not be scored, described or published as | |
| 292 | + * structured data. False covers both a post Bricks does not own and a | |
| 293 | + * Bricks page that puts `post_content` back with a Post Content element. | |
| 294 | + * | |
| 295 | + * @since 2.3.1 | |
| 296 | + * | |
| 297 | + * @param int $post_id Post being resolved. | |
| 298 | + * @return bool | |
| 299 | + */ | |
| 300 | + public static function bricks_supersedes_post_content(int $post_id): bool { | |
| 301 | + $tree = self::bricks_tree($post_id); | |
| 302 | + | |
| 303 | + if (empty($tree)) { | |
| 304 | + return false; | |
| 305 | + } | |
| 306 | + | |
| 307 | + foreach ($tree as $element) { | |
| 308 | + if (is_array($element) | |
| 309 | + && self::BRICKS_POST_CONTENT_ELEMENT === ($element['name'] ?? null) | |
| 310 | + ) { | |
| 311 | + return false; | |
| 312 | + } | |
| 313 | + } | |
| 314 | + | |
| 315 | + return !self::bricks_tree_prints_post_content($tree); | |
| 316 | + } | |
| 317 | + | |
| 318 | + /** | |
| 319 | + * Whether a Bricks tree prints the body through a dynamic-data tag. | |
| 320 | + * | |
| 321 | + * The Post Content element is not the only way back onto the page: Bricks' | |
| 322 | + * `{post_content}` tag renders the same thing from inside an ordinary text | |
| 323 | + * element, and a single-post template written that way is a common shape. | |
| 324 | + * Missing it would mean the post's real body is discarded everywhere — | |
| 325 | + * scoring, the meta/og/twitter descriptions, the schema description — for a | |
| 326 | + * page that is displaying it. | |
| 327 | + * | |
| 328 | + * Matched over the encoded tree rather than per setting, because the tag can | |
| 329 | + * sit in any string field of any element and Bricks allows modifiers after | |
| 330 | + * the name (`{post_content:...}`). | |
| 331 | + * | |
| 332 | + * @since 2.3.1 | |
| 333 | + * | |
| 334 | + * @param array $tree Bricks element tree. | |
| 335 | + * @return bool | |
| 336 | + */ | |
| 337 | + private static function bricks_tree_prints_post_content(array $tree): bool { | |
| 338 | + $encoded = wp_json_encode($tree); | |
| 339 | + | |
| 340 | + return is_string($encoded) && false !== stripos($encoded, '{post_content'); | |
| 341 | + } | |
| 342 | + | |
| 343 | + /** | |
| 344 | + * The post's content as the visitor actually receives it. | |
| 345 | + * | |
| 346 | + * `post_content` for everything except a Bricks page that discards it, and | |
| 347 | + * there the Bricks tree's text. Descriptions are derived from a post's body | |
| 348 | + * in half a dozen places; every one of them wants this rather than the raw | |
| 349 | + * column (#651). | |
| 350 | + * | |
| 351 | + * @since 2.3.1 | |
| 352 | + * | |
| 353 | + * @param \WP_Post $post Post being described. | |
| 354 | + * @return string | |
| 355 | + */ | |
| 356 | + public static function visible_content(\WP_Post $post): string { | |
| 357 | + $superseding = self::superseding_content($post); | |
| 358 | + | |
| 359 | + return '' !== $superseding ? $superseding : (string) $post->post_content; | |
| 360 | + } | |
| 361 | + | |
| 362 | + /** | |
| 363 | + * Replacement body text for a post whose `post_content` does not render. | |
| 364 | + * | |
| 365 | + * Empty for every ordinary post, which is what makes this safe to call from | |
| 366 | + * paths that already handle excerpts their own way: they keep that handling | |
| 367 | + * and only a Bricks page is diverted. | |
| 368 | + * | |
| 369 | + * @since 2.3.1 | |
| 370 | + * | |
| 371 | + * @param \WP_Post $post Post being described. | |
| 372 | + * @return string Visible body text, or '' when `post_content` is fine. | |
| 373 | + */ | |
| 374 | + public static function superseding_content(\WP_Post $post): string { | |
| 375 | + if (!self::bricks_supersedes_post_content((int) $post->ID)) { | |
| 376 | + return ''; | |
| 377 | + } | |
| 378 | + | |
| 379 | + $bricks = self::from_bricks((int) $post->ID); | |
| 380 | + | |
| 381 | + return self::is_blank($bricks) ? '' : $bricks; | |
| 382 | + } | |
| 383 | + | |
| 384 | + /** | |
| 385 | + * Body text to derive a description from, when the usual source is wrong. | |
| 386 | + * | |
| 387 | + * A hand-written excerpt is the author's own summary and is correct however | |
| 388 | + * the page is built, so it yields '' here and the caller's normal | |
| 389 | + * `get_the_excerpt()` path keeps it. Only a Bricks page with no excerpt — | |
| 390 | + * where core would derive one from discarded `post_content` — gets diverted. | |
| 391 | + * | |
| 392 | + * @since 2.3.1 | |
| 393 | + * | |
| 394 | + * @param \WP_Post $post Post being described. | |
| 395 | + * @return string Text to summarize, or '' to leave the caller's path alone. | |
| 396 | + */ | |
| 397 | + public static function superseding_excerpt_source(\WP_Post $post): string { | |
| 398 | + if ('' !== trim((string) $post->post_excerpt)) { | |
| 399 | + return ''; | |
| 400 | + } | |
| 401 | + | |
| 402 | + return self::superseding_content($post); | |
| 403 | + } | |
| 404 | + | |
| 405 | + /** | |
| 406 | + * The Bricks element tree that renders for a post. | |
| 407 | + * | |
| 408 | + * Public because what Bricks puts on the page is not only a scoring | |
| 409 | + * question: the schema graph has to know whether a Bricks element already | |
| 410 | + * publishes the page's FAQ before adding one of its own (#649, #650). | |
| 411 | + * | |
| 412 | + * Flat, in Bricks' own storage shape — `expand_bricks_components()` | |
| 413 | + * appends component definitions to the same list rather than nesting them, | |
| 414 | + * so one `foreach` reaches every element. | |
| 415 | + * | |
| 416 | + * @since 2.3.1 | |
| 417 | + * | |
| 418 | + * @param int $post_id Post being resolved. | |
| 419 | + * @return array<int,mixed> Elements, or [] when Bricks renders nothing here. | |
| 420 | + */ | |
| 421 | + /** | |
| 422 | + * The builder meta keys, for callers that need to inspect the raw storage | |
| 423 | + * rather than the text extracted from it. | |
| 424 | + * | |
| 425 | + * The SEO Analyzer reads these to answer "is there a ThinkRank FAQ element | |
| 426 | + * on this post?", which is a question about the stored tree, not about the | |
| 427 | + * words in it (#686). | |
| 428 | + * | |
| 429 | + * @since 2.7.0 | |
| 430 | + * @return string[] | |
| 431 | + */ | |
| 432 | + public static function builder_meta_keys(): array { | |
| 433 | + return self::BUILDER_META_KEYS; | |
| 434 | + } | |
| 435 | + | |
| 436 | + public static function bricks_tree(int $post_id): array { | |
| 437 | + if (array_key_exists($post_id, self::$bricks_trees)) { | |
| 438 | + return self::$bricks_trees[$post_id]; | |
| 439 | + } | |
| 440 | + | |
| 441 | + self::$bricks_trees[$post_id] = self::resolve_bricks_tree($post_id); | |
| 442 | + | |
| 443 | + return self::$bricks_trees[$post_id]; | |
| 444 | + } | |
| 445 | + | |
| 446 | + /** | |
| 447 | + * Discard the resolved-tree memo. Test seam. | |
| 448 | + * | |
| 449 | + * @since 2.3.1 | |
| 450 | + * @return void | |
| 451 | + */ | |
| 452 | + public static function flush_bricks_cache(): void { | |
| 453 | + self::$bricks_trees = []; | |
| 454 | + } | |
| 455 | + | |
| 456 | + /** | |
| 457 | + * Read and resolve a post's Bricks tree, ignoring the memo. | |
| 458 | + * | |
| 459 | + * @since 2.3.1 | |
| 460 | + * | |
| 461 | + * @param int $post_id Post being resolved. | |
| 462 | + * @return array<int,mixed> | |
| 463 | + */ | |
| 464 | + private static function resolve_bricks_tree(int $post_id): array { | |
| 465 | + if (!self::bricks_owns_post($post_id)) { | |
| 466 | + return []; | |
| 467 | + } | |
| 468 | + | |
| 469 | + $source = self::bricks_content_source($post_id); | |
| 470 | + if (!$source) { | |
| 471 | + return []; | |
| 472 | + } | |
| 473 | + | |
| 474 | + $stored = get_post_meta($source, self::bricks_meta_key(), true); | |
| 475 | + | |
| 476 | + if (is_string($stored)) { | |
| 477 | + $stored = '' === trim($stored) ? null : json_decode($stored, true); | |
| 478 | + } | |
| 479 | + | |
| 480 | + if (!is_array($stored) || empty($stored)) { | |
| 481 | + return []; | |
| 482 | + } | |
| 483 | + | |
| 484 | + return self::expand_bricks_components($stored); | |
| 485 | + } | |
| 486 | + | |
| 487 | + /** | |
| 127 | 488 | * Resolve an arbitrary chunk of editor markup for the given post. |
| 128 | 489 | * |
| 129 | 490 | * The editor sends its live content to the scorer so an author sees their |
| 130 | 491 | * unsaved edits reflected. On a builder page that live string is the raw |
| @@ -255,8 +616,350 @@ | ||
| 255 | 616 | return empty($attrs) ? '' : self::text_from_tree($attrs); |
| 256 | 617 | } |
| 257 | 618 | |
| 258 | 619 | /** |
| 620 | + * Everything Bricks contributes to this post's analyzable content. | |
| 621 | + * | |
| 622 | + * Bricks is the only builder here that needs more than a meta key, on | |
| 623 | + * three counts: | |
| 624 | + * | |
| 625 | + * - It leaves its stored tree behind when a post is switched back to the | |
| 626 | + * block editor, so an editor-mode gate has to run first or ThinkRank | |
| 627 | + * scores markup the visitor never sees — the same failure | |
| 628 | + * `_fl_builder_draft` was ordered against in #449. | |
| 629 | + * - A post's content can live on ANOTHER post. Bricks' Templates feature | |
| 630 | + * assigns a content template by condition, and a page using one stores | |
| 631 | + * nothing of its own; reading only the page's meta scores it blank | |
| 632 | + * while the visitor reads a full page. | |
| 633 | + * - Its stored text carries dynamic-data tags and internal element names | |
| 634 | + * that never reach the rendered page. | |
| 635 | + * | |
| 636 | + * @since 2.2.1 | |
| 637 | + * | |
| 638 | + * @param int $post_id Post being resolved. | |
| 639 | + * @return string Extracted text, or '' when Bricks has nothing for it. | |
| 640 | + */ | |
| 641 | + private static function from_bricks(int $post_id): string { | |
| 642 | + $tree = self::bricks_tree($post_id); | |
| 643 | + | |
| 644 | + if (empty($tree)) { | |
| 645 | + return ''; | |
| 646 | + } | |
| 647 | + | |
| 648 | + return self::strip_bricks_dynamic_tags( | |
| 649 | + self::text_from_tree(self::without_bricks_element_labels($tree)) | |
| 650 | + ); | |
| 651 | + } | |
| 652 | + | |
| 653 | + /** | |
| 654 | + * Whether Bricks — not the block editor — renders this post. | |
| 655 | + * | |
| 656 | + * Bricks writes `bricks` or `wordpress` into its editor-mode meta as the | |
| 657 | + * author toggles between the two, and never clears the content it stored | |
| 658 | + * for the other mode. Only the `wordpress` value is disqualifying: an | |
| 659 | + * absent value is the normal state for a post Bricks built and never | |
| 660 | + * toggled. This follows Bricks' own `Helpers::render_with_bricks()`, which | |
| 661 | + * bails on exactly that one value. | |
| 662 | + * | |
| 663 | + * It deliberately does not match it exactly: the comparison here is | |
| 664 | + * case-insensitive, where Bricks' is strict. Bricks 2.3.12 only ever writes | |
| 665 | + * the value lowercase, so the two agree on everything Bricks itself | |
| 666 | + * stores; they part company only on a value some other integration wrote. | |
| 667 | + * The two shipping today disagree about the casing — SureRank compares | |
| 668 | + * against `'WordPress'`, AIOSEO against `'bricks'` — and of the two ways to | |
| 669 | + * be wrong about `'WordPress'`, blocking costs a score on a page that has | |
| 670 | + * one, while allowing scores stale content the visitor never sees, which is | |
| 671 | + * the failure this gate exists to prevent. | |
| 672 | + * | |
| 673 | + * @since 2.2.1 | |
| 674 | + * | |
| 675 | + * @param int $post_id Post being resolved. | |
| 676 | + * @return bool | |
| 677 | + */ | |
| 678 | + private static function bricks_owns_post(int $post_id): bool { | |
| 679 | + $mode = get_post_meta($post_id, self::bricks_editor_mode_key(), true); | |
| 680 | + | |
| 681 | + // phpcs:ignore WordPress.WP.CapitalPDangit.MisspelledInText -- Bricks' own stored meta value, lower-cased for the comparison. | |
| 682 | + return !(is_string($mode) && 'wordpress' === strtolower(trim($mode))); | |
| 683 | + } | |
| 684 | + | |
| 685 | + /** | |
| 686 | + * The post whose Bricks tree actually renders for this post. | |
| 687 | + * | |
| 688 | + * Usually the post itself. When it stores nothing of its own, Bricks falls | |
| 689 | + * back to whichever content template's conditions match, and that template | |
| 690 | + * is a separate post carrying the words the visitor reads. | |
| 691 | + * | |
| 692 | + * Resolution is delegated to Bricks rather than reimplemented: template | |
| 693 | + * conditions are a whole rules engine (post IDs, types, taxonomies, | |
| 694 | + * archives), and a second implementation would drift from it. Bricks | |
| 695 | + * answers through statics, so they are saved and restored around the call — | |
| 696 | + * `set_active_templates()` returns early once populated, and on a | |
| 697 | + * front-end request Bricks has already populated it for the page being | |
| 698 | + * served. Clobbering that would corrupt the render in progress. | |
| 699 | + * | |
| 700 | + * Best-effort by design: any failure returns the post's own data, which is | |
| 701 | + * exactly today's behaviour. | |
| 702 | + * | |
| 703 | + * @since 2.2.1 | |
| 704 | + * | |
| 705 | + * @param int $post_id Post being resolved. | |
| 706 | + * @return int Post ID holding the Bricks tree, or 0 when there is none. | |
| 707 | + */ | |
| 708 | + private static function bricks_content_source(int $post_id): int { | |
| 709 | + $own = get_post_meta($post_id, self::bricks_meta_key(), true); | |
| 710 | + if ((is_array($own) && !empty($own)) || (is_string($own) && '' !== trim($own))) { | |
| 711 | + return $post_id; | |
| 712 | + } | |
| 713 | + | |
| 714 | + if (!class_exists('\\Bricks\\Database') | |
| 715 | + || !method_exists('\\Bricks\\Database', 'set_active_templates') | |
| 716 | + ) { | |
| 717 | + return 0; | |
| 718 | + } | |
| 719 | + | |
| 720 | + // `set_active_templates()` writes TWO statics — `$active_templates` and, | |
| 721 | + // when a header template resolves, `$header_position`. Both are saved, | |
| 722 | + // and both are restored in `finally` rather than on the happy path: a | |
| 723 | + // throw part-way through (a third-party hook on | |
| 724 | + // `bricks/database/content_type`, `bricks/builder/data_post_id` or | |
| 725 | + // `bricks/active_templates` is enough) must not leave Bricks' render | |
| 726 | + // state holding this lookup's values. Restoring only after a clean | |
| 727 | + // return is what the `catch` below would otherwise skip. | |
| 728 | + $has_header_position = property_exists('\\Bricks\\Database', 'header_position'); | |
| 729 | + $saved_templates = \Bricks\Database::$active_templates; | |
| 730 | + $saved_header_position = $has_header_position ? \Bricks\Database::$header_position : null; | |
| 731 | + | |
| 732 | + try { | |
| 733 | + \Bricks\Database::$active_templates = []; | |
| 734 | + \Bricks\Database::set_active_templates($post_id); | |
| 735 | + $template = (int) (\Bricks\Database::$active_templates['content'] ?? 0); | |
| 736 | + } catch (\Throwable $e) { | |
| 737 | + return 0; | |
| 738 | + } finally { | |
| 739 | + \Bricks\Database::$active_templates = $saved_templates; | |
| 740 | + if ($has_header_position) { | |
| 741 | + \Bricks\Database::$header_position = $saved_header_position; | |
| 742 | + } | |
| 743 | + } | |
| 744 | + | |
| 745 | + // A template that is the post itself adds nothing over the empty read | |
| 746 | + // above, and would otherwise recurse conceptually. | |
| 747 | + return $template === $post_id ? 0 : $template; | |
| 748 | + } | |
| 749 | + | |
| 750 | + /** | |
| 751 | + * Splice component definitions into the tree. | |
| 752 | + * | |
| 753 | + * A Bricks component keeps its markup in the `bricks_components` option, | |
| 754 | + * not on the page. The page stores only an instance: an element carrying | |
| 755 | + * `cid` and, usually, empty `settings`. Walking the page alone therefore | |
| 756 | + * found no words at all, and a page built entirely from components scored | |
| 757 | + * blank — the same failure as a page built from a content template. | |
| 758 | + * | |
| 759 | + * Confirmed on Bricks 2.3.12: `Bricks\Frontend::render_data()` renders the | |
| 760 | + * component's copy from an instance this walker extracted '' from. | |
| 761 | + * | |
| 762 | + * The definition is read straight from the option rather than through | |
| 763 | + * `Bricks\Helpers::get_component_instance()`. That helper resolves an | |
| 764 | + * instance's property overrides, which would be better, but it reads | |
| 765 | + * `Bricks\Database::$global_data['components']` — populated once per | |
| 766 | + * request, and empty in the admin and CLI contexts where bulk scoring | |
| 767 | + * runs. Refreshing it would mean writing to Bricks' live render state, the | |
| 768 | + * same hazard the template resolver is careful to avoid, and gating on it | |
| 769 | + * would make a page score differently in wp-admin than on the front end. | |
| 770 | + * Reading the stored definition is consistent everywhere. | |
| 771 | + * | |
| 772 | + * The trade-off: an instance that overrides a component property is scored | |
| 773 | + * with the component's authored copy rather than the override. That is the | |
| 774 | + * text the component renders by default, and it is much closer than the | |
| 775 | + * nothing this returned before. | |
| 776 | + * | |
| 777 | + * @since 2.2.1 | |
| 778 | + * | |
| 779 | + * @param array $tree Bricks content area. | |
| 780 | + * @return array Tree with component elements spliced in after each instance. | |
| 781 | + */ | |
| 782 | + private static function expand_bricks_components(array $tree): array { | |
| 783 | + $expanded = []; | |
| 784 | + $open = []; | |
| 785 | + | |
| 786 | + $walk = static function (array $elements, int $depth) use (&$walk, &$expanded, &$open): void { | |
| 787 | + foreach ($elements as $element) { | |
| 788 | + $expanded[] = $element; | |
| 789 | + | |
| 790 | + if (!is_array($element) || empty($element['cid']) || !is_string($element['cid'])) { | |
| 791 | + continue; | |
| 792 | + } | |
| 793 | + | |
| 794 | + $cid = $element['cid']; | |
| 795 | + | |
| 796 | + // A component nested inside its own definition would recurse | |
| 797 | + // forever; the depth cap covers deep but legitimate nesting. | |
| 798 | + if (isset($open[$cid]) || $depth > 4) { | |
| 799 | + continue; | |
| 800 | + } | |
| 801 | + | |
| 802 | + $children = self::bricks_component_elements($cid); | |
| 803 | + if (empty($children)) { | |
| 804 | + continue; | |
| 805 | + } | |
| 806 | + | |
| 807 | + // Re-entrant per branch, not per page: the guard is released | |
| 808 | + // after the walk so a second instance further along the page | |
| 809 | + // still expands, rather than being mistaken for recursion. | |
| 810 | + // | |
| 811 | + // That does NOT double the word count — `text_from_tree()` | |
| 812 | + // ends in `array_unique()`, which collapses a repeated | |
| 813 | + // component's copy the same way it collapses a value repeated | |
| 814 | + // across responsive breakpoints. Expanding both instances is | |
| 815 | + // about not silently dropping the second one's structure. | |
| 816 | + $open[$cid] = true; | |
| 817 | + $walk($children, $depth + 1); | |
| 818 | + unset($open[$cid]); | |
| 819 | + } | |
| 820 | + }; | |
| 821 | + | |
| 822 | + $walk($tree, 0); | |
| 823 | + | |
| 824 | + return $expanded; | |
| 825 | + } | |
| 826 | + | |
| 827 | + /** | |
| 828 | + * The stored elements of one Bricks component. | |
| 829 | + * | |
| 830 | + * @since 2.2.1 | |
| 831 | + * | |
| 832 | + * @param string $cid Component id held by an instance element. | |
| 833 | + * @return array Component elements, or [] when it cannot be resolved. | |
| 834 | + */ | |
| 835 | + private static function bricks_component_elements(string $cid): array { | |
| 836 | + $components = get_option(self::bricks_constant('BRICKS_DB_COMPONENTS', self::BRICKS_COMPONENTS_OPTION), []); | |
| 837 | + | |
| 838 | + if (!is_array($components)) { | |
| 839 | + return []; | |
| 840 | + } | |
| 841 | + | |
| 842 | + foreach ($components as $component) { | |
| 843 | + $component = self::as_children($component); | |
| 844 | + if (null === $component) { | |
| 845 | + continue; | |
| 846 | + } | |
| 847 | + | |
| 848 | + if (isset($component['id']) && $component['id'] === $cid && !empty($component['elements'])) { | |
| 849 | + return is_array($component['elements']) ? $component['elements'] : []; | |
| 850 | + } | |
| 851 | + } | |
| 852 | + | |
| 853 | + return []; | |
| 854 | + } | |
| 855 | + | |
| 856 | + /** | |
| 857 | + * Drop each Bricks element's internal name before the tree is walked. | |
| 858 | + * | |
| 859 | + * A Bricks element carries an optional top-level `label` — the nickname an | |
| 860 | + * author types in the Structure panel to find it again ("Hero headline", | |
| 861 | + * "CTA row"). It is builder chrome and is never rendered, but `label` is in | |
| 862 | + * CONTENT_KEYS because it is real content for other builders' form fields, | |
| 863 | + * so it was being counted as page copy. | |
| 864 | + * | |
| 865 | + * Only the element's own `label` is removed. A `label` inside `settings` | |
| 866 | + * is a rendered field label and stays. | |
| 867 | + * | |
| 868 | + * @since 2.2.1 | |
| 869 | + * | |
| 870 | + * @param array $tree Bricks content area. | |
| 871 | + * @return array Tree with element nicknames removed. | |
| 872 | + */ | |
| 873 | + private static function without_bricks_element_labels(array $tree): array { | |
| 874 | + foreach ($tree as $index => $element) { | |
| 875 | + if (is_array($element) && isset($element['id'], $element['label'])) { | |
| 876 | + unset($tree[$index]['label']); | |
| 877 | + } | |
| 878 | + } | |
| 879 | + | |
| 880 | + return $tree; | |
| 881 | + } | |
| 882 | + | |
| 883 | + /** | |
| 884 | + * Remove Bricks dynamic-data tags from extracted text. | |
| 885 | + * | |
| 886 | + * Bricks stores `{post_title}`, `{post_meta:price}`, `{echo:my_fn}` and the | |
| 887 | + * like verbatim and resolves them when it renders. Extraction reads the | |
| 888 | + * stored tree, so without this the placeholders were counted as words, and | |
| 889 | + * a heading whose text is `{post_title}` reported the literal token as its | |
| 890 | + * heading text. | |
| 891 | + * | |
| 892 | + * The pattern is deliberately narrower than Bricks' own | |
| 893 | + * (`/{([\wÀ-ÖØ-öø-ÿ\-\s\.\/:\(\)...]+)}/u`), which also matches braces | |
| 894 | + * containing spaces. Bricks only substitutes tags that resolve to a | |
| 895 | + * registered provider and leaves anything else on the page as literal text, | |
| 896 | + * so the broad pattern would delete prose the visitor can actually read. | |
| 897 | + * Matching only tag-shaped tokens keeps every real sentence and still | |
| 898 | + * removes every placeholder — the same trade-off SureRank makes. | |
| 899 | + * | |
| 900 | + * @since 2.2.1 | |
| 901 | + * | |
| 902 | + * @param string $text Extracted text. | |
| 903 | + * @return string Text with placeholders removed. | |
| 904 | + */ | |
| 905 | + private static function strip_bricks_dynamic_tags(string $text): string { | |
| 906 | + $stripped = preg_replace('/\{[a-z0-9_][a-z0-9_:\-\.]*\}/i', '', $text); | |
| 907 | + | |
| 908 | + if (null === $stripped) { | |
| 909 | + return $text; | |
| 910 | + } | |
| 911 | + | |
| 912 | + // Collapse the runs of spaces a removed tag leaves mid-sentence, | |
| 913 | + // without touching the newlines that separate collected nodes. | |
| 914 | + $tidied = preg_replace('/[ \t]{2,}/', ' ', $stripped); | |
| 915 | + | |
| 916 | + return null === $tidied ? $stripped : $tidied; | |
| 917 | + } | |
| 918 | + | |
| 919 | + /** | |
| 920 | + * Bricks' content-area meta key, preferring Bricks' own constant. | |
| 921 | + * | |
| 922 | + * @since 2.2.1 | |
| 923 | + * | |
| 924 | + * @return string | |
| 925 | + */ | |
| 926 | + private static function bricks_meta_key(): string { | |
| 927 | + return self::bricks_constant('BRICKS_DB_PAGE_CONTENT', self::BRICKS_CONTENT_META_KEY); | |
| 928 | + } | |
| 929 | + | |
| 930 | + /** | |
| 931 | + * Bricks' editor-mode meta key, preferring Bricks' own constant. | |
| 932 | + * | |
| 933 | + * @since 2.2.1 | |
| 934 | + * | |
| 935 | + * @return string | |
| 936 | + */ | |
| 937 | + private static function bricks_editor_mode_key(): string { | |
| 938 | + return self::bricks_constant('BRICKS_DB_EDITOR_MODE', self::BRICKS_EDITOR_MODE_META_KEY); | |
| 939 | + } | |
| 940 | + | |
| 941 | + /** | |
| 942 | + * Read one of Bricks' key-name constants, falling back to the literal. | |
| 943 | + * | |
| 944 | + * @since 2.2.1 | |
| 945 | + * | |
| 946 | + * @param string $name Constant name. | |
| 947 | + * @param string $fallback Key to use when the constant is unavailable. | |
| 948 | + * @return string | |
| 949 | + */ | |
| 950 | + private static function bricks_constant(string $name, string $fallback): string { | |
| 951 | + if (defined($name)) { | |
| 952 | + $value = constant($name); | |
| 953 | + if (is_string($value) && '' !== trim($value)) { | |
| 954 | + return $value; | |
| 955 | + } | |
| 956 | + } | |
| 957 | + | |
| 958 | + return $fallback; | |
| 959 | + } | |
| 960 | + | |
| 961 | + /** | |
| 259 | 962 | * Pull text out of whichever builder stored this post. |
| 260 | 963 | * |
| 261 | 964 | * @param int $post_id Post ID. |
| 262 | 965 | * @return string Extracted text, or '' when no builder data was found. |
| @@ -261,8 +964,15 @@ | ||
| 261 | 964 | * @param int $post_id Post ID. |
| 262 | 965 | * @return string Extracted text, or '' when no builder data was found. |
| 263 | 966 | */ |
| 264 | 967 | private static function from_builder_meta(int $post_id): string { |
| 968 | + // Bricks first: it is the only builder whose content can live on | |
| 969 | + // another post, and the only one gated on an editor mode. | |
| 970 | + $bricks = self::from_bricks($post_id); | |
| 971 | + if (!self::is_blank($bricks)) { | |
| 972 | + return $bricks; | |
| 973 | + } | |
| 974 | + | |
| 265 | 975 | foreach (self::BUILDER_META_KEYS as $key) { |
| 266 | 976 | $stored = get_post_meta($post_id, $key, true); |
| 267 | 977 | |
| 268 | 978 | if (is_string($stored) && '' !== trim($stored)) { |
| @@ -291,11 +1001,13 @@ | ||
| 291 | 1001 | |
| 292 | 1002 | continue; |
| 293 | 1003 | } |
| 294 | 1004 | |
| 295 | - // Some builders store an already-decoded array. | |
| 296 | - if (is_array($stored)) { | |
| 297 | - $text = self::text_from_tree($stored); | |
| 1005 | + // Some builders store an already-decoded tree — an array for most, | |
| 1006 | + // an array of objects for Beaver Builder (#449). | |
| 1007 | + $tree = self::as_children($stored); | |
| 1008 | + if (null !== $tree) { | |
| 1009 | + $text = self::text_from_tree($tree); | |
| 298 | 1010 | if (!self::is_blank($text)) { |
| 299 | 1011 | return $text; |
| 300 | 1012 | } |
| 301 | 1013 | } |
| @@ -304,8 +1016,43 @@ | ||
| 304 | 1016 | return ''; |
| 305 | 1017 | } |
| 306 | 1018 | |
| 307 | 1019 | /** |
| 1020 | + * A node's children, whether it stores them as an array or an object. | |
| 1021 | + * | |
| 1022 | + * The walker used to return immediately on `!is_array($node)`, so an | |
| 1023 | + * object node was dropped along with its entire subtree — silently, as | |
| 1024 | + * `''`, which the caller reads as "this builder stored nothing" rather | |
| 1025 | + * than "this walker cannot read this shape". | |
| 1026 | + * | |
| 1027 | + * Beaver Builder stores `_fl_builder_data` as an array of stdClass nodes, | |
| 1028 | + * each with a stdClass `settings` object, so every node would have been | |
| 1029 | + * dropped and adding its meta key alone would have looked like it worked | |
| 1030 | + * and changed nothing. Not BB-specific: any builder storing objects hits | |
| 1031 | + * this, and that shape will come up again (#449). | |
| 1032 | + * | |
| 1033 | + * @since 2.1.0 | |
| 1034 | + * | |
| 1035 | + * @param mixed $node Candidate node. | |
| 1036 | + * @return array<string|int,mixed>|null Traversable children, or null. | |
| 1037 | + */ | |
| 1038 | + private static function as_children($node): ?array { | |
| 1039 | + if (is_array($node)) { | |
| 1040 | + return $node; | |
| 1041 | + } | |
| 1042 | + | |
| 1043 | + // Deliberately not is_object(): a builder can store a value object | |
| 1044 | + // (DateTime, a WP_Post) whose properties are not content, and | |
| 1045 | + // get_object_vars() on those yields noise. stdClass is what the | |
| 1046 | + // JSON/serialize round-trip produces, which is the shape we want. | |
| 1047 | + if ($node instanceof \stdClass) { | |
| 1048 | + return get_object_vars($node); | |
| 1049 | + } | |
| 1050 | + | |
| 1051 | + return null; | |
| 1052 | + } | |
| 1053 | + | |
| 1054 | + /** | |
| 308 | 1055 | * Walk a builder node tree and collect the user-visible text. |
| 309 | 1056 | * |
| 310 | 1057 | * Values are joined with block-level markup so downstream heading, link and |
| 311 | 1058 | * image detection keeps working on the result. |
| @@ -325,9 +1072,10 @@ | ||
| 325 | 1072 | // happen per node rather than per leaf: a link's label and its |
| 326 | 1073 | // destination are separate sibling fields, so once the tree is |
| 327 | 1074 | // flattened to leaves the pairing is gone. |
| 328 | 1075 | $reconstruct = static function ($node) use (&$reconstruct, &$collected, &$consumed): void { |
| 329 | - if (!is_array($node)) { | |
| 1076 | + $node = self::as_children($node); | |
| 1077 | + if (null === $node) { | |
| 330 | 1078 | return; |
| 331 | 1079 | } |
| 332 | 1080 | |
| 333 | 1081 | $markup = self::markup_for_node($node, $consumed); |
| @@ -341,9 +1089,10 @@ | ||
| 341 | 1089 | // would emit the same URL a second time as a bare link, and |
| 342 | 1090 | // would turn an image's own `url` field into a spurious <a>. |
| 343 | 1091 | if (is_string($child_key) |
| 344 | 1092 | && (in_array(strtolower($child_key), self::URL_KEYS, true) |
| 345 | - || in_array(strtolower($child_key), self::IMAGE_KEYS, true)) | |
| 1093 | + || in_array(strtolower($child_key), self::IMAGE_KEYS, true) | |
| 1094 | + || in_array(strtolower($child_key), self::VIDEO_KEYS, true)) | |
| 346 | 1095 | ) { |
| 347 | 1096 | continue; |
| 348 | 1097 | } |
| 349 | 1098 | |
| @@ -353,10 +1102,11 @@ | ||
| 353 | 1102 | $reconstruct($tree); |
| 354 | 1103 | |
| 355 | 1104 | // Pass 2 — remaining visible text. |
| 356 | 1105 | $walk = static function ($node, $key = null) use (&$walk, &$collected, &$consumed): void { |
| 357 | - if (is_array($node)) { | |
| 358 | - foreach ($node as $child_key => $child) { | |
| 1106 | + $children = self::as_children($node); | |
| 1107 | + if (null !== $children) { | |
| 1108 | + foreach ($children as $child_key => $child) { | |
| 359 | 1109 | $walk($child, is_string($child_key) ? $child_key : $key); |
| 360 | 1110 | } |
| 361 | 1111 | return; |
| 362 | 1112 | } |
| @@ -394,8 +1144,79 @@ | ||
| 394 | 1144 | return implode("\n", $collected); |
| 395 | 1145 | } |
| 396 | 1146 | |
| 397 | 1147 | /** |
| 1148 | + * The video source a node is actually playing, if any. | |
| 1149 | + * | |
| 1150 | + * @since 2.3.1 | |
| 1151 | + * | |
| 1152 | + * @param array $node Builder node. | |
| 1153 | + * @return string Video source, or '' when the node carries none. | |
| 1154 | + */ | |
| 1155 | + private static function video_from(array $node): string { | |
| 1156 | + foreach ($node as $key => $value) { | |
| 1157 | + if (!is_string($key) || !is_string($value)) { | |
| 1158 | + continue; | |
| 1159 | + } | |
| 1160 | + | |
| 1161 | + if (!in_array(strtolower($key), self::VIDEO_TYPE_KEYS, true)) { | |
| 1162 | + continue; | |
| 1163 | + } | |
| 1164 | + | |
| 1165 | + $keys = self::VIDEO_KEYS_BY_TYPE[strtolower(trim($value))] ?? null; | |
| 1166 | + if (null === $keys) { | |
| 1167 | + continue; | |
| 1168 | + } | |
| 1169 | + | |
| 1170 | + // A recognised video_type settles it, including when that | |
| 1171 | + // provider's own field is empty. Falling through to the flat sweep | |
| 1172 | + // there handed back whichever sibling key happened to come first in | |
| 1173 | + // node order — the stale youtube_url left behind after switching | |
| 1174 | + // the widget to a hosted file, which is exactly what keying on the | |
| 1175 | + // declared type is meant to prevent. | |
| 1176 | + $declared = self::url_from($node, $keys); | |
| 1177 | + | |
| 1178 | + return self::is_video_source($declared) ? $declared : ''; | |
| 1179 | + } | |
| 1180 | + | |
| 1181 | + $url = self::url_from($node, self::VIDEO_KEYS); | |
| 1182 | + | |
| 1183 | + return self::is_video_source($url) ? $url : ''; | |
| 1184 | + } | |
| 1185 | + | |
| 1186 | + /** | |
| 1187 | + * Whether a value can be a video source. | |
| 1188 | + * | |
| 1189 | + * `looks_like_url()` also accepts `#anchor`, `mailto:` and `tel:`, which a | |
| 1190 | + * link node may legitimately hold but a video cannot: `<iframe src="#top">` | |
| 1191 | + * is not a video and would reach a video sitemap as one. | |
| 1192 | + * | |
| 1193 | + * @since 2.3.1 | |
| 1194 | + * | |
| 1195 | + * @param string $url Candidate source. | |
| 1196 | + * @return bool | |
| 1197 | + */ | |
| 1198 | + private static function is_video_source(string $url): bool { | |
| 1199 | + return '' !== $url | |
| 1200 | + && (1 === preg_match('#^(https?:)?//#i', $url) || str_starts_with($url, '/')); | |
| 1201 | + } | |
| 1202 | + | |
| 1203 | + /** | |
| 1204 | + * Whether a video source points at a file rather than a provider page. | |
| 1205 | + * | |
| 1206 | + * @since 2.3.1 | |
| 1207 | + * | |
| 1208 | + * @param string $url Video source. | |
| 1209 | + * @return bool | |
| 1210 | + */ | |
| 1211 | + private static function is_video_file(string $url): bool { | |
| 1212 | + $path = (string) wp_parse_url($url, PHP_URL_PATH); | |
| 1213 | + $ext = strtolower((string) pathinfo($path, PATHINFO_EXTENSION)); | |
| 1214 | + | |
| 1215 | + return in_array($ext, self::VIDEO_FILE_EXTENSIONS, true); | |
| 1216 | + } | |
| 1217 | + | |
| 1218 | + /** | |
| 398 | 1219 | * Rebuild the HTML a single builder node represents, if any. |
| 399 | 1220 | * |
| 400 | 1221 | * Looks only at the node's own fields (plus one level of nesting, because |
| 401 | 1222 | * builders commonly wrap a destination as `{ url: … }`). Returns an empty |
| @@ -411,12 +1232,23 @@ | ||
| 411 | 1232 | private static function markup_for_node(array $node, array &$consumed): string { |
| 412 | 1233 | $text = self::first_value($node, self::CONTENT_KEYS); |
| 413 | 1234 | $url = self::url_from($node, self::URL_KEYS); |
| 414 | 1235 | $image = self::image_from($node); |
| 1236 | + $video = self::video_from($node); | |
| 415 | 1237 | $tag = self::heading_tag_from($node); |
| 416 | 1238 | |
| 417 | 1239 | $parts = []; |
| 418 | 1240 | |
| 1241 | + // Video: an embed shape rather than a link, so the video detector can | |
| 1242 | + // see it while the link counters do not mistake it for an outbound | |
| 1243 | + // link. A file source becomes <video src>, anything else an <iframe>, | |
| 1244 | + // matching how the builder itself renders the two cases. | |
| 1245 | + if ('' !== $video) { | |
| 1246 | + $parts[] = self::is_video_file($video) | |
| 1247 | + ? sprintf('<video src="%s"></video>', esc_url_raw($video)) | |
| 1248 | + : sprintf('<iframe src="%s"></iframe>', esc_url_raw($video)); | |
| 1249 | + } | |
| 1250 | + | |
| 419 | 1251 | // Image: alt text matters as much as the tag, since alt checks run over |
| 420 | 1252 | // whatever this returns. |
| 421 | 1253 | if ('' !== $image['url']) { |
| 422 | 1254 | $alt = '' !== $image['alt'] ? $image['alt'] : (string) self::first_value($node, self::ALT_KEYS); |
| @@ -490,10 +1322,11 @@ | ||
| 490 | 1322 | return trim($value); |
| 491 | 1323 | } |
| 492 | 1324 | |
| 493 | 1325 | // Elementor and Breakdance both nest the destination one level down. |
| 494 | - if (is_array($value)) { | |
| 495 | - foreach ($value as $nested_key => $nested) { | |
| 1326 | + $nested_values = self::as_children($value); | |
| 1327 | + if (null !== $nested_values) { | |
| 1328 | + foreach ($nested_values as $nested_key => $nested) { | |
| 496 | 1329 | if (is_string($nested_key) |
| 497 | 1330 | && in_array(strtolower($nested_key), ['url', 'href', 'permalink'], true) |
| 498 | 1331 | && is_string($nested) |
| 499 | 1332 | && self::looks_like_url($nested) |
| @@ -522,12 +1355,13 @@ | ||
| 522 | 1355 | if (is_string($value) && self::looks_like_url($value)) { |
| 523 | 1356 | return ['url' => trim($value), 'alt' => '']; |
| 524 | 1357 | } |
| 525 | 1358 | |
| 526 | - if (is_array($value)) { | |
| 1359 | + $nested_values = self::as_children($value); | |
| 1360 | + if (null !== $nested_values) { | |
| 527 | 1361 | $url = ''; |
| 528 | 1362 | $alt = ''; |
| 529 | - foreach ($value as $nested_key => $nested) { | |
| 1363 | + foreach ($nested_values as $nested_key => $nested) { | |
| 530 | 1364 | if (!is_string($nested_key) || !is_string($nested)) { |
| 531 | 1365 | continue; |
| 532 | 1366 | } |
| 533 | 1367 | $nested_key = strtolower($nested_key); |
| @@ -603,13 +1437,30 @@ | ||
| 603 | 1437 | return (bool) preg_match('#^(mailto:|tel:|\#)#i', $value); |
| 604 | 1438 | } |
| 605 | 1439 | |
| 606 | 1440 | /** |
| 607 | - * Whether a value carries no readable text. | |
| 1441 | + * Whether a value carries nothing worth analyzing. | |
| 608 | 1442 | * |
| 1443 | + * Readable text is the usual signal, but not the only one: a page can be | |
| 1444 | + * made entirely of media. A builder section holding just a gallery | |
| 1445 | + * reconstructs to `<img>` tags and one holding just a video widget to a | |
| 1446 | + * single `<iframe>` — both strip to an empty string, so a text-only test | |
| 1447 | + * discarded them here and the page fell through to the next builder key, | |
| 1448 | + * then to the raw markup, and finally reported as having no content at all. | |
| 1449 | + * | |
| 1450 | + * Comments are dropped before the tag test: the raw markup this class falls | |
| 1451 | + * back to on a builder page is unrendered block comments, which must stay | |
| 1452 | + * blank rather than be mistaken for reconstructed media. | |
| 1453 | + * | |
| 609 | 1454 | * @param string $value Candidate content. |
| 610 | 1455 | * @return bool |
| 611 | 1456 | */ |
| 612 | 1457 | private static function is_blank(string $value): bool { |
| 613 | - return '' === trim(wp_strip_all_tags($value)); | |
| 1458 | + if ('' !== trim(wp_strip_all_tags($value))) { | |
| 1459 | + return false; | |
| 1460 | + } | |
| 1461 | + | |
| 1462 | + $without_comments = (string) preg_replace('~<!--.*?-->~s', '', $value); | |
| 1463 | + | |
| 1464 | + return 1 !== preg_match('~<(?:a|img|iframe|video|source)\b~i', $without_comments); | |
| 614 | 1465 | } |
| 615 | 1466 | } |
