| @@ -87,23 +87,34 @@ | ||
| 87 | 87 | 'Book', 'Movie', 'Service', 'ImageObject', 'VideoObject', |
| 88 | 88 | ]; |
| 89 | 89 | |
| 90 | 90 | /** |
| 91 | - * Gutenberg FAQ block name. | |
| 91 | + * Types that may carry a `breadcrumb` property. | |
| 92 | + * | |
| 93 | + * Schema.org limits `breadcrumb` to WebPage and its subtypes. Attaching it | |
| 94 | + * to a BlogPosting, Product or Recipe primary fails validation with | |
| 95 | + * "Unexpected property" on every URL; the standalone BreadcrumbList node is | |
| 96 | + * emitted either way (#693). | |
| 97 | + * | |
| 98 | + * @since 2.7.0 | |
| 99 | + * @var array<int,string> | |
| 92 | 100 | */ |
| 93 | - private const FAQ_BLOCK = 'thinkrank/faq'; | |
| 101 | + private const BREADCRUMB_TYPES = [ | |
| 102 | + 'WebPage', 'AboutPage', 'CheckoutPage', 'CollectionPage', 'ContactPage', | |
| 103 | + 'FAQPage', 'ItemPage', 'MedicalWebPage', 'ProfilePage', 'QAPage', | |
| 104 | + 'RealEstateListing', 'SearchResultsPage', 'MediaGallery', 'ImageGallery', | |
| 105 | + 'VideoGallery', | |
| 106 | + ]; | |
| 94 | 107 | |
| 95 | 108 | /** |
| 96 | - * Elementor FAQ widget name. | |
| 97 | - */ | |
| 98 | - private const FAQ_WIDGET = 'thinkrank-faq'; | |
| 99 | - | |
| 100 | - /** | |
| 101 | - * Bricks FAQ element name. | |
| 109 | + * The FAQ surfaces themselves are read by FAQ_Content, which owns their | |
| 110 | + * names, their stored shapes and their per-surface schema toggles. Only the | |
| 111 | + * *foreign* FAQ producers are listed below, because standing down for them | |
| 112 | + * is this class's decision rather than a fact about where ThinkRank keeps | |
| 113 | + * its own content (#767). | |
| 102 | 114 | * |
| 103 | - * @since 2.3.1 | |
| 115 | + * @since 2.10.1 | |
| 104 | 116 | */ |
| 105 | - private const FAQ_BRICKS_ELEMENT = 'thinkrank-faq'; | |
| 106 | 117 | |
| 107 | 118 | /** |
| 108 | 119 | * Third-party Elementor widgets that publish their own FAQPage. |
| 109 | 120 | * |
| @@ -139,8 +150,16 @@ | ||
| 139 | 150 | */ |
| 140 | 151 | private static ?self $instance = null; |
| 141 | 152 | |
| 142 | 153 | /** |
| 154 | + * Memoised master switch, or null when it has not been read this request. | |
| 155 | + * | |
| 156 | + * @since 2.7.0 | |
| 157 | + * @var bool|null | |
| 158 | + */ | |
| 159 | + private static ?bool $master_switch_on = null; | |
| 160 | + | |
| 161 | + /** | |
| 143 | 162 | * Competing page-level entities: ['rank' => int, 'schema' => array, 'type' => string]. |
| 144 | 163 | * |
| 145 | 164 | * @var array<int,array> |
| 146 | 165 | */ |
| @@ -222,8 +241,10 @@ | ||
| 222 | 241 | * @return void |
| 223 | 242 | */ |
| 224 | 243 | public static function reset(): void { |
| 225 | 244 | self::$instance = null; |
| 245 | + // Or a test that seeds the switch inherits the previous test's answer. | |
| 246 | + self::$master_switch_on = null; | |
| 226 | 247 | } |
| 227 | 248 | |
| 228 | 249 | /** |
| 229 | 250 | * Register a candidate for the page's single page-level entity. |
| @@ -292,8 +313,61 @@ | ||
| 292 | 313 | return (is_string($actual) && $actual !== '') ? $actual : $declared; |
| 293 | 314 | } |
| 294 | 315 | |
| 295 | 316 | /** |
| 317 | + * Whether a page entity may carry a `breadcrumb` property. | |
| 318 | + * | |
| 319 | + * Reading the node's own `@type` key is not enough, and every case it | |
| 320 | + * misses ends with a real WebPage losing a valid property: | |
| 321 | + * | |
| 322 | + * - A deployed schema whose stored JSON omits `@type` carries the type in | |
| 323 | + * the `schema_type` column instead. `effective_type()` already resolves | |
| 324 | + * that, which is why the node's `@id` reads `#webpage` even though the | |
| 325 | + * node itself has no `@type` — so the resolved value is what has to be | |
| 326 | + * consulted here too. | |
| 327 | + * - JSON-LD permits several types on one node. `["WebPage", "FAQPage"]` is | |
| 328 | + * a WebPage, but a strict in_array() against the array as a whole is | |
| 329 | + * false, so the trail would be dropped from a page that may carry it. | |
| 330 | + * - A list naming nothing usable (`[]`, `[null]`) is the first case wearing | |
| 331 | + * the second's clothes, and resolves the same way. | |
| 332 | + * | |
| 333 | + * @since 2.7.0 | |
| 334 | + * @param array $node The page entity. | |
| 335 | + * @param string $resolved_type Type the graph resolved for it. | |
| 336 | + * @return bool | |
| 337 | + */ | |
| 338 | + private function allows_breadcrumb(array $node, string $resolved_type): bool { | |
| 339 | + $declared = $node['@type'] ?? ''; | |
| 340 | + | |
| 341 | + // One path for both shapes. Splitting them invites the list branch to | |
| 342 | + // grow its own idea of what an absent type means, which is the mistake | |
| 343 | + // being corrected here in the first place. | |
| 344 | + $named = false; | |
| 345 | + | |
| 346 | + foreach (is_array($declared) ? $declared : [$declared] as $type) { | |
| 347 | + if (!is_string($type) || '' === $type) { | |
| 348 | + continue; | |
| 349 | + } | |
| 350 | + | |
| 351 | + $named = true; | |
| 352 | + | |
| 353 | + if (in_array($type, self::BREADCRUMB_TYPES, true)) { | |
| 354 | + return true; | |
| 355 | + } | |
| 356 | + } | |
| 357 | + | |
| 358 | + // The node named a type, and none of them may carry a breadcrumb. | |
| 359 | + if ($named) { | |
| 360 | + return false; | |
| 361 | + } | |
| 362 | + | |
| 363 | + // It named none, so it is whatever the graph resolved for it — the same | |
| 364 | + // value its @id was minted from. An empty list is no more informative | |
| 365 | + // than a missing key and must not read as "definitely not a WebPage". | |
| 366 | + return in_array($resolved_type, self::BREADCRUMB_TYPES, true); | |
| 367 | + } | |
| 368 | + | |
| 369 | + /** | |
| 296 | 370 | * Register a node that does not compete for the page-level slot. |
| 297 | 371 | * |
| 298 | 372 | * @since 1.32.0 |
| 299 | 373 | * @param array $schema Schema array. |
| @@ -474,201 +548,151 @@ | ||
| 474 | 548 | // page throws `post_content` away, so a FAQ block left there when the |
| 475 | 549 | // page was switched over never renders. Publishing its questions would |
| 476 | 550 | // put schema on the page for content no visitor can see — which Google |
| 477 | 551 | // treats as a violation, not merely a duplicate (#650). |
| 478 | - if (!$this->bricks_supersedes_post_content((int) $post->ID)) { | |
| 479 | - $this->collect_block_faq($post); | |
| 480 | - } | |
| 552 | + foreach (self::faq_groups($post) as $group) { | |
| 553 | + // The stored toggle decides whether a producer contributes schema. | |
| 554 | + // FAQ_Content reports it rather than applying it, because a block | |
| 555 | + // with schema switched off is still visible FAQ content that the | |
| 556 | + // abilities have to describe (#767). | |
| 557 | + if (empty($group['schema'])) { | |
| 558 | + continue; | |
| 559 | + } | |
| 481 | 560 | |
| 482 | - $this->collect_elementor_faq($post); | |
| 483 | - $this->collect_bricks_faq($post); | |
| 484 | - } | |
| 485 | - | |
| 486 | - /** | |
| 487 | - * Whether Bricks renders this post and discards its `post_content`. | |
| 488 | - * | |
| 489 | - * @since 2.3.1 | |
| 490 | - * @param int $post_id Post being viewed. | |
| 491 | - * @return bool | |
| 492 | - */ | |
| 493 | - private function bricks_supersedes_post_content(int $post_id): bool { | |
| 494 | - if (!class_exists('ThinkRank\\SEO\\Builder_Content')) { | |
| 495 | - $file = THINKRANK_PLUGIN_DIR . 'includes/seo/class-builder-content.php'; | |
| 496 | - if (!file_exists($file)) { | |
| 497 | - return false; | |
| 498 | - } | |
| 499 | - require_once $file; | |
| 561 | + $this->absorb_content_faq($this->questions_from_pairs($group['pairs'])); | |
| 500 | 562 | } |
| 501 | - | |
| 502 | - return \ThinkRank\SEO\Builder_Content::bricks_supersedes_post_content($post_id); | |
| 503 | 563 | } |
| 504 | 564 | |
| 505 | 565 | /** |
| 506 | - * Record that a body FAQ producer's content is represented in the graph. | |
| 566 | + * Whether ThinkRank would publish a FAQPage for this post. | |
| 507 | 567 | * |
| 508 | - * Deliberately not keyed on the entity count growing: when a block asks the | |
| 509 | - * same question as the per-post deployment, dedup means nothing is added, | |
| 510 | - * but the block's content *is* covered and it must still stay quiet. | |
| 568 | + * Answers the question `get-faq` has to put to an agent — "is this actually | |
| 569 | + * being emitted?" — without rendering the page. Kept here rather than in the | |
| 570 | + * ability because every term of it is this class's own rule: the password | |
| 571 | + * form, the master switch, the per-content-type Schema switch, the other | |
| 572 | + * plugins whose FAQ output makes ThinkRank stand down, and the filter that | |
| 573 | + * overrides all of it. | |
| 511 | 574 | * |
| 512 | - * @since 1.32.0 | |
| 513 | - * @param array $entities Questions found on that producer. | |
| 514 | - * @return void | |
| 515 | - */ | |
| 516 | - private function absorb_content_faq(array $entities): void { | |
| 517 | - if (empty($entities)) { | |
| 518 | - return; | |
| 519 | - } | |
| 520 | - | |
| 521 | - $this->add_faq_entities($entities); | |
| 522 | - $this->absorbed_content_faq = true; | |
| 523 | - } | |
| 524 | - | |
| 525 | - /** | |
| 526 | - * Collect FAQ questions from thinkrank/faq blocks, including nested ones. | |
| 575 | + * It answers for the post as a visitor gets it. A draft or a scheduled post | |
| 576 | + * is reported on what it will emit once it is served, because that is the | |
| 577 | + * question an agent preparing one is asking; a password-protected post is | |
| 578 | + * not, because publishing it changes nothing — the form stays. | |
| 527 | 579 | * |
| 528 | - * @since 1.32.0 | |
| 529 | - * @param \WP_Post $post Post being viewed. | |
| 530 | - * @return void | |
| 580 | + * @since 2.10.1 | |
| 581 | + * @param \WP_Post $post Post to test. | |
| 582 | + * @return bool | |
| 531 | 583 | */ |
| 532 | - private function collect_block_faq(\WP_Post $post): void { | |
| 533 | - if (!function_exists('parse_blocks') || !has_blocks($post->post_content)) { | |
| 534 | - return; | |
| 584 | + public static function will_emit_faqpage(\WP_Post $post): bool { | |
| 585 | + // The same gate collect_post_faq() applies, and for the same reason: | |
| 586 | + // behind a password form the block never renders, so no FAQPage is | |
| 587 | + // published. Without this the ability answered "yes, it is emitted" | |
| 588 | + // about a page that emits nothing — which is the answer that stops an | |
| 589 | + // agent looking any further. | |
| 590 | + if (function_exists('post_password_required') && post_password_required($post)) { | |
| 591 | + return false; | |
| 535 | 592 | } |
| 536 | 593 | |
| 537 | - $this->walk_blocks(parse_blocks($post->post_content)); | |
| 538 | - } | |
| 594 | + $has_questions = false; | |
| 539 | 595 | |
| 540 | - /** | |
| 541 | - * Recurse a parsed block tree collecting FAQ entries. | |
| 542 | - * | |
| 543 | - * @since 1.32.0 | |
| 544 | - * @param array $blocks Parsed blocks. | |
| 545 | - * @return void | |
| 546 | - */ | |
| 547 | - private function walk_blocks(array $blocks): void { | |
| 548 | - foreach ($blocks as $block) { | |
| 549 | - if (!is_array($block)) { | |
| 596 | + foreach (self::faq_groups($post) as $group) { | |
| 597 | + if (empty($group['schema'])) { | |
| 550 | 598 | continue; |
| 551 | 599 | } |
| 552 | 600 | |
| 553 | - if (($block['blockName'] ?? '') === self::FAQ_BLOCK) { | |
| 554 | - $attrs = $block['attrs'] ?? []; | |
| 555 | - | |
| 556 | - // Mirrors Blocks_Manager: schema is on unless explicitly disabled. | |
| 557 | - $disabled = array_key_exists('outputSchema', $attrs) && false === $attrs['outputSchema']; | |
| 558 | - | |
| 559 | - if (!$disabled) { | |
| 560 | - $this->absorb_content_faq($this->questions_from_pairs($attrs['faqs'] ?? [])); | |
| 601 | + foreach ($group['pairs'] as $pair) { | |
| 602 | + if ('' !== trim((string) ($pair['question'] ?? '')) && '' !== trim((string) ($pair['answer'] ?? ''))) { | |
| 603 | + $has_questions = true; | |
| 604 | + break 2; | |
| 561 | 605 | } |
| 562 | 606 | } |
| 607 | + } | |
| 563 | 608 | |
| 564 | - if (!empty($block['innerBlocks']) && is_array($block['innerBlocks'])) { | |
| 565 | - $this->walk_blocks($block['innerBlocks']); | |
| 566 | - } | |
| 609 | + if (!$has_questions) { | |
| 610 | + return false; | |
| 567 | 611 | } |
| 612 | + | |
| 613 | + if (!self::schema_allowed_for_post($post)) { | |
| 614 | + return false; | |
| 615 | + } | |
| 616 | + | |
| 617 | + $emit = !(new self())->has_foreign_faq_source($post); | |
| 618 | + | |
| 619 | + /** This filter is documented in includes/frontend/class-schema-graph.php */ | |
| 620 | + return (bool) apply_filters('thinkrank_emit_faqpage', $emit, $post); | |
| 568 | 621 | } |
| 569 | 622 | |
| 570 | 623 | /** |
| 571 | - * Collect FAQ questions from Elementor FAQ widgets. | |
| 624 | + * output_allowed(), asked about a named post instead of the current query. | |
| 572 | 625 | * |
| 573 | - * @since 1.32.0 | |
| 574 | - * @param \WP_Post $post Post being viewed. | |
| 575 | - * @return void | |
| 626 | + * output_allowed() resolves the content type from the main query, which in | |
| 627 | + * an admin or MCP request is not the post being asked about. | |
| 628 | + * | |
| 629 | + * @since 2.10.1 | |
| 630 | + * @param \WP_Post $post Post to test. | |
| 631 | + * @return bool | |
| 576 | 632 | */ |
| 577 | - private function collect_elementor_faq(\WP_Post $post): void { | |
| 578 | - $raw = get_post_meta($post->ID, '_elementor_data', true); | |
| 579 | - if (empty($raw) || !is_string($raw)) { | |
| 580 | - return; | |
| 633 | + private static function schema_allowed_for_post(\WP_Post $post): bool { | |
| 634 | + if (null === self::$master_switch_on) { | |
| 635 | + // Resolve the site-wide half through the existing reader. | |
| 636 | + self::output_allowed(); | |
| 581 | 637 | } |
| 582 | 638 | |
| 583 | - $elements = json_decode($raw, true); | |
| 584 | - if (!is_array($elements)) { | |
| 585 | - return; | |
| 639 | + if (!self::$master_switch_on) { | |
| 640 | + return false; | |
| 586 | 641 | } |
| 587 | 642 | |
| 588 | - $this->walk_elementor($elements); | |
| 589 | - } | |
| 643 | + if (class_exists('ThinkRank\\SEO\\Content_Type_Settings')) { | |
| 644 | + return \ThinkRank\SEO\Content_Type_Settings::is_enabled( | |
| 645 | + \ThinkRank\SEO\Content_Type_Settings::FEATURE_SCHEMA, | |
| 646 | + (string) $post->post_type, | |
| 647 | + true | |
| 648 | + ); | |
| 649 | + } | |
| 590 | 650 | |
| 591 | - /** | |
| 592 | - * Collect FAQ questions from Bricks FAQ elements. | |
| 593 | - * | |
| 594 | - * Reads the tree Bricks will actually render — resolved through | |
| 595 | - * `Builder_Content`, so a page whose content lives on a content template or | |
| 596 | - * inside a component is covered, and one switched back to the block editor | |
| 597 | - * is not. | |
| 598 | - * | |
| 599 | - * Unlike the block, this is not gated on Bricks owning `post_content`: a | |
| 600 | - * Bricks element is on the page whenever Bricks renders the page, which is | |
| 601 | - * exactly what resolving the tree already establishes (#626). | |
| 602 | - * | |
| 603 | - * @since 2.3.1 | |
| 604 | - * @param \WP_Post $post Post being viewed. | |
| 605 | - * @return void | |
| 606 | - */ | |
| 607 | - private function collect_bricks_faq(\WP_Post $post): void { | |
| 608 | - $this->walk_bricks($this->bricks_tree((int) $post->ID)); | |
| 651 | + return true; | |
| 609 | 652 | } |
| 610 | 653 | |
| 611 | 654 | /** |
| 612 | - * Collect FAQ entries from a resolved Bricks tree. | |
| 655 | + * Every FAQ producer on a post, loaded defensively. | |
| 613 | 656 | * |
| 614 | - * The tree is flat, so no recursion: `Builder_Content::bricks_tree()` | |
| 615 | - * splices component definitions into the same list. | |
| 657 | + * The reader lives in the SEO namespace, and this class runs in contexts | |
| 658 | + * where that autoloader is not guaranteed — which is why the Bricks gate it | |
| 659 | + * replaced carried the same require. | |
| 616 | 660 | * |
| 617 | - * The element's own settings are read here rather than through | |
| 618 | - * `FAQ_Element`, whose class extends `Bricks\Element` and so cannot even be | |
| 619 | - * loaded when the theme is inactive — which is exactly the case that still | |
| 620 | - * has a stored tree, on a site that has since switched themes. The repeater | |
| 621 | - * uses the same `question` / `answer` keys as the block, so the shared | |
| 622 | - * builder below already understands it. | |
| 623 | - * | |
| 624 | - * @since 2.3.1 | |
| 625 | - * @param array $elements Bricks elements. | |
| 626 | - * @return void | |
| 661 | + * @since 2.10.1 | |
| 662 | + * @param \WP_Post $post Post being read. | |
| 663 | + * @return array<int, array{source: string, schema: bool, pairs: array}> | |
| 627 | 664 | */ |
| 628 | - private function walk_bricks(array $elements): void { | |
| 629 | - foreach ($elements as $element) { | |
| 630 | - if (!is_array($element) || ($element['name'] ?? '') !== self::FAQ_BRICKS_ELEMENT) { | |
| 631 | - continue; | |
| 665 | + private static function faq_groups(\WP_Post $post): array { | |
| 666 | + if (!class_exists('ThinkRank\\SEO\\FAQ_Content')) { | |
| 667 | + $file = THINKRANK_PLUGIN_DIR . 'includes/seo/class-faq-content.php'; | |
| 668 | + if (!file_exists($file)) { | |
| 669 | + return []; | |
| 632 | 670 | } |
| 671 | + require_once $file; | |
| 672 | + } | |
| 633 | 673 | |
| 634 | - $settings = is_array($element['settings'] ?? null) ? $element['settings'] : []; | |
| 635 | - | |
| 636 | - // Mirrors FAQ_Element: a cleared Bricks checkbox loses its key. | |
| 637 | - if (empty($settings['outputSchema'])) { | |
| 638 | - continue; | |
| 639 | - } | |
| 640 | - | |
| 641 | - $this->absorb_content_faq($this->questions_from_pairs($settings['faqs'] ?? [])); | |
| 642 | - } | |
| 674 | + return \ThinkRank\SEO\FAQ_Content::groups($post); | |
| 643 | 675 | } |
| 644 | 676 | |
| 645 | 677 | /** |
| 646 | - * Recurse an Elementor element tree collecting FAQ entries. | |
| 678 | + * Record that a body FAQ producer's content is represented in the graph. | |
| 647 | 679 | * |
| 680 | + * Deliberately not keyed on the entity count growing: when a block asks the | |
| 681 | + * same question as the per-post deployment, dedup means nothing is added, | |
| 682 | + * but the block's content *is* covered and it must still stay quiet. | |
| 683 | + * | |
| 648 | 684 | * @since 1.32.0 |
| 649 | - * @param array $elements Elementor elements. | |
| 685 | + * @param array $entities Questions found on that producer. | |
| 650 | 686 | * @return void |
| 651 | 687 | */ |
| 652 | - private function walk_elementor(array $elements): void { | |
| 653 | - foreach ($elements as $element) { | |
| 654 | - if (!is_array($element)) { | |
| 655 | - continue; | |
| 656 | - } | |
| 688 | + private function absorb_content_faq(array $entities): void { | |
| 689 | + if (empty($entities)) { | |
| 690 | + return; | |
| 691 | + } | |
| 657 | 692 | |
| 658 | - if (($element['widgetType'] ?? '') === self::FAQ_WIDGET) { | |
| 659 | - $settings = $element['settings'] ?? []; | |
| 660 | - | |
| 661 | - // Mirrors FAQ_Widget: schema unless the toggle is off. | |
| 662 | - if ('yes' === ($settings['output_schema'] ?? 'yes')) { | |
| 663 | - $this->absorb_content_faq($this->questions_from_pairs($settings['faqs'] ?? [])); | |
| 664 | - } | |
| 665 | - } | |
| 666 | - | |
| 667 | - if (!empty($element['elements']) && is_array($element['elements'])) { | |
| 668 | - $this->walk_elementor($element['elements']); | |
| 669 | - } | |
| 670 | - } | |
| 693 | + $this->add_faq_entities($entities); | |
| 694 | + $this->absorbed_content_faq = true; | |
| 671 | 695 | } |
| 672 | 696 | |
| 673 | 697 | /** |
| 674 | 698 | * Turn stored question/answer pairs into Question entities. |
| @@ -727,8 +751,59 @@ | ||
| 727 | 751 | return !empty($this->primary_candidates) || !empty($this->supporting) || !empty($this->faq_entities); |
| 728 | 752 | } |
| 729 | 753 | |
| 730 | 754 | /** |
| 755 | + * Whether ThinkRank may emit structured data for this request at all. | |
| 756 | + * | |
| 757 | + * The master switch and the matrix's per-content-type Schema switch, asked | |
| 758 | + * once. #461 put the master switch inside output_site_schema_markup(), | |
| 759 | + * which is one of four producers; the other three never learned about it, | |
| 760 | + * so turning Schema off removed the deployed rows and left the live | |
| 761 | + * generator running — the page emitted *more* types with the switch off | |
| 762 | + * than with it on (#688). | |
| 763 | + * | |
| 764 | + * Static because the producers that need it do not share a base class: the | |
| 765 | + * three graph producers converge on render(), but Blocks_Manager emits its | |
| 766 | + * own script tag from a content filter and never touches the graph, so it | |
| 767 | + * has to ask the same question independently. | |
| 768 | + * | |
| 769 | + * @since 2.7.0 | |
| 770 | + * @return bool True when structured data may be emitted. | |
| 771 | + */ | |
| 772 | + public static function output_allowed(): bool { | |
| 773 | + // Memoised because inject_block_schema() asks once per matching block, | |
| 774 | + // and Schema_Management_System's constructor builds a schema builder and | |
| 775 | + // a cache manager and registers listeners — it is not something to spin | |
| 776 | + // up per block. The switch is site-wide, so it cannot change within a | |
| 777 | + // request; the per-content-type check below is query-dependent and stays | |
| 778 | + // live. | |
| 779 | + if (null === self::$master_switch_on) { | |
| 780 | + self::$master_switch_on = true; | |
| 781 | + | |
| 782 | + if (class_exists('ThinkRank\\SEO\\Schema_Management_System')) { | |
| 783 | + $settings = (new \ThinkRank\SEO\Schema_Management_System())->get_settings('site', null); | |
| 784 | + | |
| 785 | + // Absent means "not configured", which every other reader treats | |
| 786 | + // as enabled; only a value that is present and off disables. | |
| 787 | + self::$master_switch_on = !(array_key_exists('enabled', $settings) && empty($settings['enabled'])); | |
| 788 | + } | |
| 789 | + } | |
| 790 | + | |
| 791 | + if (!self::$master_switch_on) { | |
| 792 | + return false; | |
| 793 | + } | |
| 794 | + | |
| 795 | + if (class_exists('ThinkRank\\SEO\\Content_Type_Settings')) { | |
| 796 | + return \ThinkRank\SEO\Content_Type_Settings::is_enabled_for_current( | |
| 797 | + \ThinkRank\SEO\Content_Type_Settings::FEATURE_SCHEMA, | |
| 798 | + true | |
| 799 | + ); | |
| 800 | + } | |
| 801 | + | |
| 802 | + return true; | |
| 803 | + } | |
| 804 | + | |
| 805 | + /** | |
| 731 | 806 | * Assemble and emit the graph. Safe to call more than once. |
| 732 | 807 | * |
| 733 | 808 | * @since 1.32.0 |
| 734 | 809 | * @return void |
| @@ -737,8 +812,14 @@ | ||
| 737 | 812 | if ($this->rendered || !$this->has_nodes()) { |
| 738 | 813 | return; |
| 739 | 814 | } |
| 740 | 815 | |
| 816 | + // Every graph producer converges here, so this is the one place the | |
| 817 | + // master switch has to hold for all of them (#688). | |
| 818 | + if (!self::output_allowed()) { | |
| 819 | + return; | |
| 820 | + } | |
| 821 | + | |
| 741 | 822 | // A 404 response represents no content, so there is nothing for |
| 742 | 823 | // structured data to describe. The page-level producers already skip |
| 743 | 824 | // this context, but the site-identity entity does not, so without this |
| 744 | 825 | // guard every miss — including crawlers probing URLs that never existed |
| @@ -773,8 +854,18 @@ | ||
| 773 | 854 | if (empty($graph)) { |
| 774 | 855 | return; |
| 775 | 856 | } |
| 776 | 857 | |
| 858 | + // One pass over the assembled graph, rather than at each producer. | |
| 859 | + // @id and url values arrive from a dozen of them — some derived from | |
| 860 | + // WordPress, some read straight out of stored settings — and on a | |
| 861 | + // misconfigured site that produced a single graph carrying both | |
| 862 | + // schemes at once, with @ids that no longer matched the canonical they | |
| 863 | + // are supposed to identify (#638). Normalizing where the graph is | |
| 864 | + // serialized is the only place that catches all of them, including | |
| 865 | + // nodes an add-on added through the filter above. | |
| 866 | + $graph = \ThinkRank\SEO\Url_Scheme::apply_deep($graph); | |
| 867 | + | |
| 777 | 868 | $json = wp_json_encode( |
| 778 | 869 | ['@context' => self::SCHEMA_CONTEXT, '@graph' => array_values($graph)], |
| 779 | 870 | JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT |
| 780 | 871 | | JSON_HEX_TAG | JSON_HEX_AMP | JSON_HEX_APOS | JSON_HEX_QUOT |
| @@ -784,9 +875,9 @@ | ||
| 784 | 875 | return; |
| 785 | 876 | } |
| 786 | 877 | |
| 787 | 878 | echo "<!-- ThinkRank Schema Graph -->\n"; |
| 788 | - echo '<script type="application/ld+json">' . "\n"; | |
| 879 | + echo '<script type="application/ld+json" data-thinkrank="schema-graph">' . "\n"; | |
| 789 | 880 | echo $json . "\n"; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- wp_json_encode with JSON_HEX_* cannot break out of the script block. |
| 790 | 881 | echo '</script>' . "\n"; |
| 791 | 882 | echo "<!-- /ThinkRank Schema Graph -->\n"; |
| 792 | 883 | } |
| @@ -897,11 +988,12 @@ | ||
| 897 | 988 | $primary = ['schema' => $faq, 'type' => 'FAQPage']; |
| 898 | 989 | $faq = null; |
| 899 | 990 | } |
| 900 | 991 | |
| 901 | - $nodes = []; | |
| 902 | - $primary_id = ''; | |
| 903 | - $used_ids = []; | |
| 992 | + $nodes = []; | |
| 993 | + $primary_id = ''; | |
| 994 | + $primary_type = ''; | |
| 995 | + $used_ids = []; | |
| 904 | 996 | |
| 905 | 997 | if (null !== $primary) { |
| 906 | 998 | $node = $primary['schema']; |
| 907 | 999 | |
| @@ -909,9 +1001,10 @@ | ||
| 909 | 1001 | // so an "Article" setting that renders BlogPosting reads #blogposting. |
| 910 | 1002 | $resolved_type = $this->effective_type($node, $primary['type']); |
| 911 | 1003 | |
| 912 | 1004 | $node = $this->assign_id($node, $base . '#' . strtolower($resolved_type), $used_ids); |
| 913 | - $primary_id = $node['@id']; | |
| 1005 | + $primary_id = $node['@id']; | |
| 1006 | + $primary_type = $resolved_type; | |
| 914 | 1007 | $nodes['primary'] = $node; |
| 915 | 1008 | } |
| 916 | 1009 | |
| 917 | 1010 | // Entities deployed alongside the winner (Pro's Multi-Schema lets a post |
| @@ -952,8 +1045,9 @@ | ||
| 952 | 1045 | $person_nodes = []; |
| 953 | 1046 | |
| 954 | 1047 | foreach ($this->supporting as $index => $node) { |
| 955 | 1048 | $type = $node['@type'] ?? ''; |
| 1049 | + $slot = $this->site_level_slot($type); | |
| 956 | 1050 | |
| 957 | 1051 | if ('BreadcrumbList' === $type) { |
| 958 | 1052 | $node = $this->assign_id($node, $base . '#breadcrumb', $used_ids); |
| 959 | 1053 | $breadcrumb_id = $node['@id']; |
| @@ -962,9 +1056,9 @@ | ||
| 962 | 1056 | $website_id = $node['@id']; |
| 963 | 1057 | } elseif ('Organization' === $type) { |
| 964 | 1058 | $node = $this->assign_id($node, home_url('/#organization'), $used_ids); |
| 965 | 1059 | $organization_nodes[] = $node; |
| 966 | - } elseif (in_array($type, self::SITE_LEVEL_TYPES, true)) { | |
| 1060 | + } elseif ('' !== $slot) { | |
| 967 | 1061 | // Site-level entities describe the site, not the page, so their |
| 968 | 1062 | // @id must be stable across URLs. Falling through to the |
| 969 | 1063 | // page-scoped branch minted a fresh identity on every URL, so |
| 970 | 1064 | // one business became N entities in a crawler's graph and |
| @@ -974,9 +1068,9 @@ | ||
| 974 | 1068 | // arrive here claiming the same @id. assign_id() would resolve |
| 975 | 1069 | // that collision by minting "#person-2", turning a duplicate |
| 976 | 1070 | // into two competing entities that split the identity a |
| 977 | 1071 | // knowledge graph is meant to consolidate (#479). |
| 978 | - $duplicate_key = $this->find_same_entity($nodes, $type, $node); | |
| 1072 | + $duplicate_key = $this->find_same_entity($nodes, $slot, $node); | |
| 979 | 1073 | |
| 980 | 1074 | if (null !== $duplicate_key) { |
| 981 | 1075 | $nodes[$duplicate_key] = $this->merge_entity($nodes[$duplicate_key], $node); |
| 982 | 1076 | continue; |
| @@ -981,11 +1075,11 @@ | ||
| 981 | 1075 | $nodes[$duplicate_key] = $this->merge_entity($nodes[$duplicate_key], $node); |
| 982 | 1076 | continue; |
| 983 | 1077 | } |
| 984 | 1078 | |
| 985 | - $node = $this->assign_id($node, home_url('/#' . strtolower($type)), $used_ids); | |
| 1079 | + $node = $this->assign_id($node, home_url('/#' . strtolower($slot)), $used_ids); | |
| 986 | 1080 | |
| 987 | - if ('Person' === $type) { | |
| 1081 | + if ('Person' === $slot) { | |
| 988 | 1082 | $person_nodes[] = $node; |
| 989 | 1083 | } |
| 990 | 1084 | } elseif (is_string($type) && $type !== '') { |
| 991 | 1085 | $node = $this->assign_id($node, $base . '#' . strtolower($type), $used_ids); |
| @@ -998,9 +1092,13 @@ | ||
| 998 | 1092 | if (isset($nodes['primary'])) { |
| 999 | 1093 | if ($website_id !== '' && !isset($nodes['primary']['isPartOf'])) { |
| 1000 | 1094 | $nodes['primary']['isPartOf'] = ['@id' => $website_id]; |
| 1001 | 1095 | } |
| 1002 | - if ($breadcrumb_id !== '' && !isset($nodes['primary']['breadcrumb'])) { | |
| 1096 | + if ( | |
| 1097 | + $breadcrumb_id !== '' | |
| 1098 | + && !isset($nodes['primary']['breadcrumb']) | |
| 1099 | + && $this->allows_breadcrumb($nodes['primary'], $primary_type) | |
| 1100 | + ) { | |
| 1003 | 1101 | $nodes['primary']['breadcrumb'] = ['@id' => $breadcrumb_id]; |
| 1004 | 1102 | } |
| 1005 | 1103 | |
| 1006 | 1104 | // Point publisher/author at the full nodes already in the graph. |
| @@ -1074,8 +1172,32 @@ | ||
| 1074 | 1172 | return ['winner' => array_shift($kept), 'siblings' => array_values($kept)]; |
| 1075 | 1173 | } |
| 1076 | 1174 | |
| 1077 | 1175 | /** |
| 1176 | + * The site-level entity a node's @type makes it, or '' for none. | |
| 1177 | + * | |
| 1178 | + * A LocalBusiness is published under the subtype the site chose in Local | |
| 1179 | + * SEO ("Dentist", "Restaurant"), but it is still the one business entity, | |
| 1180 | + * so every subtype shares the LocalBusiness slot: the same home-scoped | |
| 1181 | + * `/#localbusiness` @id it always had, and the same entity for | |
| 1182 | + * find_same_entity() to fold a per-post copy into. Matching the literal | |
| 1183 | + * type instead would have given a Dentist a fresh page-scoped @id on every | |
| 1184 | + * URL, the exact split #471 closed. | |
| 1185 | + * | |
| 1186 | + * @since 2.10.0 | |
| 1187 | + * | |
| 1188 | + * @param mixed $type Node @type. | |
| 1189 | + * @return string 'LocalBusiness', 'Person' or ''. | |
| 1190 | + */ | |
| 1191 | + private function site_level_slot($type): string { | |
| 1192 | + if (\ThinkRank\Config\Local_Business_Types_Config::is_local_business($type)) { | |
| 1193 | + return 'LocalBusiness'; | |
| 1194 | + } | |
| 1195 | + | |
| 1196 | + return (is_string($type) && in_array($type, self::SITE_LEVEL_TYPES, true)) ? $type : ''; | |
| 1197 | + } | |
| 1198 | + | |
| 1199 | + /** | |
| 1078 | 1200 | * Find an already-placed node describing the same entity as $node. |
| 1079 | 1201 | * |
| 1080 | 1202 | * Identity is `email` when both carry one — two people can share a name, |
| 1081 | 1203 | * but not a mailbox — and a case-insensitive `name` match otherwise. A node |
| @@ -1084,9 +1206,9 @@ | ||
| 1084 | 1206 | * |
| 1085 | 1207 | * @since 2.0.2 |
| 1086 | 1208 | * |
| 1087 | 1209 | * @param array $nodes Nodes placed so far, keyed. |
| 1088 | - * @param string $type Schema type to match within. | |
| 1210 | + * @param string $type Site-level slot to match within (see site_level_slot()). | |
| 1089 | 1211 | * @param array $node Candidate node. |
| 1090 | 1212 | * @return string|null Key of the matching node, or null. |
| 1091 | 1213 | */ |
| 1092 | 1214 | private function find_same_entity(array $nodes, string $type, array $node): ?string { |
| @@ -1097,9 +1219,9 @@ | ||
| 1097 | 1219 | return null; |
| 1098 | 1220 | } |
| 1099 | 1221 | |
| 1100 | 1222 | foreach ($nodes as $key => $placed) { |
| 1101 | - if (($placed['@type'] ?? '') !== $type) { | |
| 1223 | + if ($this->site_level_slot($placed['@type'] ?? '') !== $type) { | |
| 1102 | 1224 | continue; |
| 1103 | 1225 | } |
| 1104 | 1226 | |
| 1105 | 1227 | $placed_email = isset($placed['email']) ? strtolower(trim((string) $placed['email'])) : ''; |
