| @@ -14,8 +14,13 @@ | ||
| 14 | 14 | declare(strict_types=1); |
| 15 | 15 | |
| 16 | 16 | namespace ThinkRank\SEO; |
| 17 | 17 | |
| 18 | +// Prevent direct access | |
| 19 | +if (!defined('ABSPATH')) { | |
| 20 | + exit; | |
| 21 | +} | |
| 22 | + | |
| 18 | 23 | // Ensure dependencies are loaded |
| 19 | 24 | if (!class_exists('ThinkRank\\SEO\\Abstract_SEO_Manager')) { |
| 20 | 25 | require_once THINKRANK_PLUGIN_DIR . 'includes/seo/class-abstract-seo-manager.php'; |
| 21 | 26 | } |
| @@ -60,8 +65,18 @@ | ||
| 60 | 65 | */ |
| 61 | 66 | private string $last_delivery_warning = ''; |
| 62 | 67 | |
| 63 | 68 | /** |
| 69 | + * Whether the last save's delivery-mode switch failed outright, as opposed | |
| 70 | + * to succeeding with a warning. Both set {@see delivery_switch_warning()}, | |
| 71 | + * and only one of them means /llms.txt is still on the old path. | |
| 72 | + * | |
| 73 | + * @since 2.1.0 | |
| 74 | + * @var bool | |
| 75 | + */ | |
| 76 | + private bool $last_delivery_switch_failed = false; | |
| 77 | + | |
| 78 | + /** | |
| 64 | 79 | * LLMs.txt content sections configuration |
| 65 | 80 | * |
| 66 | 81 | * @since 1.0.0 |
| 67 | 82 | * @var array |
| @@ -75,9 +90,9 @@ | ||
| 75 | 90 | ], |
| 76 | 91 | 'key_features' => [ |
| 77 | 92 | 'title' => 'Key Features', |
| 78 | 93 | 'required' => true, |
| 79 | - 'description' => 'Main features and functionality of the website', | |
| 94 | + 'description' => 'Main features and functionality of the website, one feature per line. Commas are part of a feature, not separators.', | |
| 80 | 95 | 'max_length' => 300 |
| 81 | 96 | ], |
| 82 | 97 | 'architecture' => [ |
| 83 | 98 | 'title' => 'Architecture & Components', |
| @@ -148,8 +163,40 @@ | ||
| 148 | 163 | */ |
| 149 | 164 | private const PUBLISHED_AT_OPTION = 'thinkrank_llms_txt_published_at'; |
| 150 | 165 | |
| 151 | 166 | /** |
| 167 | + * Option recording what the site's public URL really answers /llms.txt with. | |
| 168 | + * | |
| 169 | + * Shaped as ['home' => string, 'result' => 'charset'|'no_charset'|'unknown', | |
| 170 | + * 'checked_at' => int] and keyed on the home URL, so a clone or a migration | |
| 171 | + * re-checks instead of inheriting the verdict of the host it came from. | |
| 172 | + * | |
| 173 | + * @since 2.1.0 | |
| 174 | + * @var string | |
| 175 | + */ | |
| 176 | + private const DELIVERY_PROBE_OPTION = 'thinkrank_llms_delivery_probe'; | |
| 177 | + | |
| 178 | + /** | |
| 179 | + * How long an inconclusive delivery check is left alone before retrying. | |
| 180 | + * | |
| 181 | + * A conclusive verdict stands until the document is published again; only | |
| 182 | + * the "could not tell" answer — a blocked loopback, an HTTP-auth'd staging | |
| 183 | + * site — is worth asking about a second time, and not often. | |
| 184 | + * | |
| 185 | + * @since 2.1.0 | |
| 186 | + * @var int | |
| 187 | + */ | |
| 188 | + private const DELIVERY_PROBE_RETRY = DAY_IN_SECONDS; | |
| 189 | + | |
| 190 | + /** | |
| 191 | + * Shown when the server answers the published file without a charset. | |
| 192 | + * | |
| 193 | + * @since 2.1.0 | |
| 194 | + * @var string | |
| 195 | + */ | |
| 196 | + private const STATIC_CHARSET_WARNING = 'This server answers the published llms.txt without a character set, so accented characters and curly quotes arrive mis-decoded. Set Delivery Method to "Served by WordPress" to publish it as UTF-8.'; | |
| 197 | + | |
| 198 | + /** | |
| 152 | 199 | * Delivery modes accepted by the `delivery_mode` setting. |
| 153 | 200 | * |
| 154 | 201 | * @since 2.1.0 |
| 155 | 202 | * @var string[] |
| @@ -156,8 +203,20 @@ | ||
| 156 | 203 | */ |
| 157 | 204 | private const DELIVERY_MODES = ['auto', 'static', 'dynamic']; |
| 158 | 205 | |
| 159 | 206 | /** |
| 207 | + * One-time marker for {@see LLMs_Txt_Manager::maybe_migrate_legacy_key_features()}. | |
| 208 | + * | |
| 209 | + * Public so the activator can record it on a fresh install, which has no | |
| 210 | + * value saved under the old comma rule and must never be migrated. | |
| 211 | + * | |
| 212 | + * @since 2.10.0 | |
| 213 | + * @var string | |
| 214 | + */ | |
| 215 | + public const KEY_FEATURES_MIGRATION_OPTION = 'thinkrank_llms_key_features_migration'; | |
| 216 | + public const KEY_FEATURES_MIGRATION_VERSION = '1'; | |
| 217 | + | |
| 218 | + /** | |
| 160 | 219 | * Business type templates for content generation |
| 161 | 220 | * |
| 162 | 221 | * @since 1.0.0 |
| 163 | 222 | * @var array |
| @@ -249,8 +308,13 @@ | ||
| 249 | 308 | 'validation' => [], |
| 250 | 309 | 'file_info' => [] |
| 251 | 310 | ]; |
| 252 | 311 | |
| 312 | + // Generating from saved settings (the MCP ability passes an empty | |
| 313 | + // payload) can happen before any admin request has run the upgrade, | |
| 314 | + // so make sure a legacy comma list has been converted first. | |
| 315 | + self::maybe_migrate_legacy_key_features(); | |
| 316 | + | |
| 253 | 317 | // Get current settings |
| 254 | 318 | $settings = $this->get_settings('site'); |
| 255 | 319 | |
| 256 | 320 | // Merge saved settings underneath the provided input so that empty or |
| @@ -317,8 +381,9 @@ | ||
| 317 | 381 | */ |
| 318 | 382 | public function save_settings(string $context_type, ?int $context_id, array $settings): bool { |
| 319 | 383 | $this->last_unpublish_failed = false; |
| 320 | 384 | $this->last_delivery_warning = ''; |
| 385 | + $this->last_delivery_switch_failed = false; | |
| 321 | 386 | $previous_mode = $this->resolve_delivery_mode(); |
| 322 | 387 | |
| 323 | 388 | $result = parent::save_settings($context_type, $context_id, $settings); |
| 324 | 389 | |
| @@ -412,9 +477,18 @@ | ||
| 412 | 477 | if (empty($write['success'])) { |
| 413 | 478 | $this->last_delivery_warning = isset($write['message']) && '' !== (string) $write['message'] |
| 414 | 479 | ? (string) $write['message'] |
| 415 | 480 | : 'The delivery method was saved, but the published llms.txt could not be moved to it.'; |
| 481 | + $this->last_delivery_switch_failed = true; | |
| 482 | + return; | |
| 416 | 483 | } |
| 484 | + | |
| 485 | + // The switch worked, but static delivery on this server cannot carry the | |
| 486 | + // charset the document needs. Only an explicitly chosen `static` gets | |
| 487 | + // this far — `auto` moves itself to WordPress delivery instead. | |
| 488 | + if (!empty($write['delivery_warning'])) { | |
| 489 | + $this->last_delivery_warning = (string) $write['delivery_warning']; | |
| 490 | + } | |
| 417 | 491 | } |
| 418 | 492 | |
| 419 | 493 | /** |
| 420 | 494 | * Message from the last save whose delivery-mode switch could not be |
| @@ -428,8 +502,20 @@ | ||
| 428 | 502 | return $this->last_delivery_warning; |
| 429 | 503 | } |
| 430 | 504 | |
| 431 | 505 | /** |
| 506 | + * Whether the last save's warning was a failed switch rather than a | |
| 507 | + * successful one the server cannot serve correctly. | |
| 508 | + * | |
| 509 | + * @since 2.1.0 | |
| 510 | + * | |
| 511 | + * @return bool | |
| 512 | + */ | |
| 513 | + public function delivery_switch_failed(): bool { | |
| 514 | + return $this->last_delivery_switch_failed; | |
| 515 | + } | |
| 516 | + | |
| 517 | + /** | |
| 432 | 518 | * Whether the last save_settings() disabled the feature but could not remove |
| 433 | 519 | * the published llms.txt file (which may therefore still be served). |
| 434 | 520 | * |
| 435 | 521 | * @return bool |
| @@ -447,11 +533,16 @@ | ||
| 447 | 533 | * Apache/LiteSpeed, the stacks that read the .htaccess charset block — on |
| 448 | 534 | * nginx a physical file is served with a bare `Content-Type: text/plain` |
| 449 | 535 | * that neither fix path can reach, which renders UTF-8 as mojibake (#419). |
| 450 | 536 | * |
| 451 | - * Layered hosts (e.g. an nginx front end reporting as something else) can | |
| 452 | - * defeat the detection, which is why the setting also accepts an explicit | |
| 453 | - * override rather than relying on $is_apache alone. | |
| 537 | + * $is_apache is not trusted on its own: WordPress reads it from | |
| 538 | + * $_SERVER['SERVER_SOFTWARE'], which describes the server that runs PHP | |
| 539 | + * rather than the one answering the public request. A reverse proxy hides | |
| 540 | + * the difference — an nginx edge in front of an Apache backend reports | |
| 541 | + * Apache, so `auto` chose the file that nginx then served with no charset, | |
| 542 | + * which is the very defect the setting was added to avoid (#493). A | |
| 543 | + * publish-time self-request settles what the detection cannot see, and its | |
| 544 | + * verdict is what this consults; an explicit setting still wins outright. | |
| 454 | 545 | * |
| 455 | 546 | * @since 2.1.0 |
| 456 | 547 | * |
| 457 | 548 | * @param string|null $mode Optional. Raw setting value; read from the saved |
| @@ -467,13 +558,214 @@ | ||
| 467 | 558 | if ('static' === $mode || 'dynamic' === $mode) { |
| 468 | 559 | return $mode; |
| 469 | 560 | } |
| 470 | 561 | |
| 562 | + // A root PHP cannot write to has no static path at all: publishing | |
| 563 | + // would simply fail and /llms.txt would 404. The sitemap's `auto` | |
| 564 | + // already resolves this way (#754); llms.txt did not, so on an | |
| 565 | + // Apache/LiteSpeed host with a read-only root — a managed stack such as | |
| 566 | + // Flywheel, where ABSPATH is the locked core folder — `auto` chose | |
| 567 | + // static and then could not deliver it (#756). | |
| 568 | + if (!wp_is_writable(ABSPATH)) { | |
| 569 | + return 'dynamic'; | |
| 570 | + } | |
| 571 | + | |
| 471 | 572 | // $is_apache also covers LiteSpeed, which reads .htaccess the same way. |
| 472 | - return !empty($GLOBALS['is_apache']) ? 'static' : 'dynamic'; | |
| 573 | + if (empty($GLOBALS['is_apache'])) { | |
| 574 | + return 'dynamic'; | |
| 575 | + } | |
| 576 | + | |
| 577 | + // Detection says this stack reads the .htaccess charset block. Believe | |
| 578 | + // it unless a self-request has caught the public URL answering without | |
| 579 | + // a charset, which is what a reverse-proxied stack does (#493). | |
| 580 | + return $this->static_delivery_drops_charset() ? 'dynamic' : 'static'; | |
| 473 | 581 | } |
| 474 | 582 | |
| 475 | 583 | /** |
| 584 | + * Whether the recorded check caught the public URL dropping the charset. | |
| 585 | + * | |
| 586 | + * @since 2.1.0 | |
| 587 | + * | |
| 588 | + * @return bool | |
| 589 | + */ | |
| 590 | + private function static_delivery_drops_charset(): bool { | |
| 591 | + return 'no_charset' === ($this->delivery_probe()['result'] ?? ''); | |
| 592 | + } | |
| 593 | + | |
| 594 | + /** | |
| 595 | + * The delivery check recorded for this site, or [] when there is none. | |
| 596 | + * | |
| 597 | + * @since 2.1.0 | |
| 598 | + * | |
| 599 | + * @return array | |
| 600 | + */ | |
| 601 | + private function delivery_probe(): array { | |
| 602 | + $probe = get_option(self::DELIVERY_PROBE_OPTION, []); | |
| 603 | + | |
| 604 | + if (!is_array($probe) || !isset($probe['result'])) { | |
| 605 | + return []; | |
| 606 | + } | |
| 607 | + | |
| 608 | + return ($probe['home'] ?? '') === home_url() ? $probe : []; | |
| 609 | + } | |
| 610 | + | |
| 611 | + /** | |
| 612 | + * Ask the site's own public URL what it answers /llms.txt with. | |
| 613 | + * | |
| 614 | + * @since 2.1.0 | |
| 615 | + * | |
| 616 | + * @return string 'charset', 'no_charset', or 'unknown' when the response | |
| 617 | + * could not be read and nothing should be concluded from it. | |
| 618 | + */ | |
| 619 | + private function probe_static_delivery(): string { | |
| 620 | + if (!function_exists('wp_remote_get')) { | |
| 621 | + return 'unknown'; | |
| 622 | + } | |
| 623 | + | |
| 624 | + // The cache-buster stops a page cache from answering with a copy stored | |
| 625 | + // before the file was written; a server ignores the query string when it | |
| 626 | + // serves a physical file, so the response still shows the real headers. | |
| 627 | + $url = add_query_arg( | |
| 628 | + 'thinkrank-delivery-check', | |
| 629 | + (string) time(), | |
| 630 | + home_url('/llms.txt') | |
| 631 | + ); | |
| 632 | + | |
| 633 | + $response = wp_remote_get($url, [ | |
| 634 | + 'timeout' => 5, | |
| 635 | + 'redirection' => 2, | |
| 636 | + // A request to our own home URL, from which a single response header | |
| 637 | + // is read. Staging and local installs routinely run on certificates | |
| 638 | + // this host does not trust, and failing there would leave the very | |
| 639 | + // sites most likely to be misconfigured unchecked. | |
| 640 | + 'sslverify' => false, | |
| 641 | + 'headers' => ['Cache-Control' => 'no-cache'], | |
| 642 | + ]); | |
| 643 | + | |
| 644 | + if (is_wp_error($response) || 200 !== (int) wp_remote_retrieve_response_code($response)) { | |
| 645 | + return 'unknown'; | |
| 646 | + } | |
| 647 | + | |
| 648 | + $content_type = wp_remote_retrieve_header($response, 'content-type'); | |
| 649 | + | |
| 650 | + // A header sent more than once comes back as an array. | |
| 651 | + if (is_array($content_type)) { | |
| 652 | + $content_type = implode(' ', $content_type); | |
| 653 | + } | |
| 654 | + | |
| 655 | + $content_type = trim((string) $content_type); | |
| 656 | + | |
| 657 | + // No Content-Type at all is the same problem: the browser is left to | |
| 658 | + // guess the encoding. | |
| 659 | + if ('' === $content_type) { | |
| 660 | + return 'no_charset'; | |
| 661 | + } | |
| 662 | + | |
| 663 | + return false !== stripos($content_type, 'charset=') ? 'charset' : 'no_charset'; | |
| 664 | + } | |
| 665 | + | |
| 666 | + /** | |
| 667 | + * Persist the outcome of a delivery check. | |
| 668 | + * | |
| 669 | + * @since 2.1.0 | |
| 670 | + * | |
| 671 | + * @param string $verdict One of 'charset', 'no_charset', 'unknown'. | |
| 672 | + * @return void | |
| 673 | + */ | |
| 674 | + private function record_delivery_probe(string $verdict): void { | |
| 675 | + update_option(self::DELIVERY_PROBE_OPTION, [ | |
| 676 | + 'home' => home_url(), | |
| 677 | + 'result' => $verdict, | |
| 678 | + 'checked_at' => time(), | |
| 679 | + ], false); | |
| 680 | + } | |
| 681 | + | |
| 682 | + /** | |
| 683 | + * Confirm the published file is really served with a charset, and act on it. | |
| 684 | + * | |
| 685 | + * Static delivery leans on an .htaccess directive, so it is only ever as | |
| 686 | + * good as the guess that the server reads .htaccess. This checks the guess | |
| 687 | + * against the response the public URL actually returns: a site left on | |
| 688 | + * `auto` is moved to WordPress delivery when the charset is missing — the | |
| 689 | + * file has to go with it, or it would shadow the PHP route that carries the | |
| 690 | + * charset — while a site that asked for `static` keeps its file and gets a | |
| 691 | + * warning, because an explicit choice is not overruled. | |
| 692 | + * | |
| 693 | + * @since 2.1.0 | |
| 694 | + * | |
| 695 | + * @param array $result Publish result to annotate. | |
| 696 | + * @return array The annotated result. | |
| 697 | + */ | |
| 698 | + private function verify_static_delivery(array $result): array { | |
| 699 | + $verdict = $this->probe_static_delivery(); | |
| 700 | + | |
| 701 | + $this->record_delivery_probe($verdict); | |
| 702 | + | |
| 703 | + if ('no_charset' !== $verdict) { | |
| 704 | + return $result; | |
| 705 | + } | |
| 706 | + | |
| 707 | + $settings = $this->get_settings('site'); | |
| 708 | + $explicit = 'static' === (string) ($settings['delivery_mode'] ?? 'auto'); | |
| 709 | + | |
| 710 | + // Auto: resolve_delivery_mode() answers 'dynamic' from here on, so the | |
| 711 | + // file it would otherwise leave behind has to be removed. The document | |
| 712 | + // is already stored, so nothing is lost by deleting it. | |
| 713 | + if (!$explicit && $this->delete_static_file()) { | |
| 714 | + delete_transient('thinkrank_llms_file_status'); | |
| 715 | + $this->purge_llms_txt_caches(); | |
| 716 | + | |
| 717 | + $result['delivery_mode'] = 'dynamic'; | |
| 718 | + $result['charset_pinned'] = true; | |
| 719 | + $result['message'] = 'LLMs.txt published. This server answers a static file without a character set, so WordPress serves it as UTF-8 instead.'; | |
| 720 | + $result['permissions']['file_exists'] = false; | |
| 721 | + $result['permissions']['file_writable'] = null; | |
| 722 | + | |
| 723 | + return $result; | |
| 724 | + } | |
| 725 | + | |
| 726 | + $result['charset_pinned'] = false; | |
| 727 | + $result['delivery_warning'] = self::STATIC_CHARSET_WARNING; | |
| 728 | + $result['message'] = trim((string) $result['message'] . ' ' . self::STATIC_CHARSET_WARNING); | |
| 729 | + | |
| 730 | + return $result; | |
| 731 | + } | |
| 732 | + | |
| 733 | + /** | |
| 734 | + * Whether the delivery check may run on this request. | |
| 735 | + * | |
| 736 | + * It makes an HTTP request of its own, so it never runs on a front-end | |
| 737 | + * page view — only where an administrator, the REST API, WP-CLI or cron is | |
| 738 | + * already waiting on a status read. | |
| 739 | + * | |
| 740 | + * @since 2.1.0 | |
| 741 | + * | |
| 742 | + * @return bool | |
| 743 | + */ | |
| 744 | + private function delivery_probe_is_due(): bool { | |
| 745 | + $interactive = is_admin() | |
| 746 | + || (defined('REST_REQUEST') && REST_REQUEST) | |
| 747 | + || (defined('WP_CLI') && WP_CLI) | |
| 748 | + || (function_exists('wp_doing_cron') && wp_doing_cron()); | |
| 749 | + | |
| 750 | + if (!$interactive) { | |
| 751 | + return false; | |
| 752 | + } | |
| 753 | + | |
| 754 | + $probe = $this->delivery_probe(); | |
| 755 | + | |
| 756 | + if ([] === $probe) { | |
| 757 | + return true; | |
| 758 | + } | |
| 759 | + | |
| 760 | + if ('unknown' !== $probe['result']) { | |
| 761 | + return false; | |
| 762 | + } | |
| 763 | + | |
| 764 | + return (time() - (int) ($probe['checked_at'] ?? 0)) > self::DELIVERY_PROBE_RETRY; | |
| 765 | + } | |
| 766 | + | |
| 767 | + /** | |
| 476 | 768 | * The published llms.txt document, or an empty string when unpublished. |
| 477 | 769 | * |
| 478 | 770 | * @since 2.1.0 |
| 479 | 771 | * |
| @@ -485,8 +777,28 @@ | ||
| 485 | 777 | return is_string($content) ? $content : ''; |
| 486 | 778 | } |
| 487 | 779 | |
| 488 | 780 | /** |
| 781 | + * Whether /llms.txt is currently being served, in either delivery mode. | |
| 782 | + * | |
| 783 | + * `static` publishes a file at ABSPATH; `dynamic` keeps the document in | |
| 784 | + * an option and answers from serve_llms_txt(). Callers that only need | |
| 785 | + * this yes/no must use it in preference to get_llms_txt_status(), which | |
| 786 | + * resolves the delivery mode, may fire a loopback delivery probe, asks | |
| 787 | + * the filesystem API whether ABSPATH is writable, reads the document and | |
| 788 | + * writes a transient — far too much work for a boolean, and not | |
| 789 | + * something a dashboard summary should be triggering. | |
| 790 | + * | |
| 791 | + * @since 2.2.1 | |
| 792 | + * | |
| 793 | + * @return bool | |
| 794 | + */ | |
| 795 | + public function is_published(): bool { | |
| 796 | + return file_exists(ABSPATH . 'llms.txt') | |
| 797 | + || '' !== trim($this->get_published_content()); | |
| 798 | + } | |
| 799 | + | |
| 800 | + /** | |
| 489 | 801 | * Ask the common page/CDN cache layers to drop their copy of /llms.txt. |
| 490 | 802 | * |
| 491 | 803 | * A cached response outlives a republish, so without this a mode switch or |
| 492 | 804 | * a content change keeps serving the old document (and, on the static path, |
| @@ -535,13 +847,33 @@ | ||
| 535 | 847 | * |
| 536 | 848 | * @return bool True once nothing is left to serve. |
| 537 | 849 | */ |
| 538 | 850 | public function delete_llms_txt_file(): bool { |
| 539 | - delete_transient('thinkrank_llms_file_status'); | |
| 540 | - | |
| 541 | 851 | delete_option(self::CONTENT_OPTION); |
| 542 | 852 | delete_option(self::PUBLISHED_AT_OPTION); |
| 543 | 853 | |
| 854 | + return $this->unpublish_static_file(); | |
| 855 | + } | |
| 856 | + | |
| 857 | + /** | |
| 858 | + * Stop serving llms.txt, but keep the document. | |
| 859 | + * | |
| 860 | + * Deactivation needs this half: the physical file must go — it shadows the | |
| 861 | + * next plugin's routes and advertises a plugin that is switched off — but | |
| 862 | + * the user's prose has to survive so reactivation can republish it. | |
| 863 | + * {@see \ThinkRank\Core\Activator::restore_webroot_artifacts()} does that. | |
| 864 | + * | |
| 865 | + * Deactivation previously called {@see delete_llms_txt_file()}, which drops | |
| 866 | + * the stored document too, so a deactivate/reactivate round-trip silently | |
| 867 | + * lost whatever the user had written. | |
| 868 | + * | |
| 869 | + * @since 2.1.0 | |
| 870 | + * | |
| 871 | + * @return bool True once nothing is left on disk. | |
| 872 | + */ | |
| 873 | + public function unpublish_static_file(): bool { | |
| 874 | + delete_transient('thinkrank_llms_file_status'); | |
| 875 | + | |
| 544 | 876 | $removed = $this->delete_static_file(); |
| 545 | 877 | $this->purge_llms_txt_caches(); |
| 546 | 878 | |
| 547 | 879 | return $removed; |
| @@ -828,9 +1160,12 @@ | ||
| 828 | 1160 | // Keep the stored copy in step with the file so a later switch to |
| 829 | 1161 | // dynamic delivery serves the same document. |
| 830 | 1162 | $this->store_published_content($content); |
| 831 | 1163 | |
| 832 | - return $result; | |
| 1164 | + // Detection said this server reads the .htaccess block. Check what the | |
| 1165 | + // public URL really answers with before leaving the file in place — on a | |
| 1166 | + // reverse-proxied stack the detection describes the wrong server (#493). | |
| 1167 | + return $this->verify_static_delivery($result); | |
| 833 | 1168 | } |
| 834 | 1169 | |
| 835 | 1170 | /** |
| 836 | 1171 | * Persist the published document and bust the caches that mirror it. |
| @@ -869,8 +1204,19 @@ | ||
| 869 | 1204 | } |
| 870 | 1205 | |
| 871 | 1206 | $llms_file = ABSPATH . 'llms.txt'; |
| 872 | 1207 | $mode = $this->resolve_delivery_mode(); |
| 1208 | + | |
| 1209 | + // A site that published before this check existed — or whose server has | |
| 1210 | + // changed under it — has never had its delivery confirmed. Do it here so | |
| 1211 | + // an already-broken install heals without waiting for a republish; the | |
| 1212 | + // recorded verdict and the status cache keep it to a couple of requests | |
| 1213 | + // a day at most. | |
| 1214 | + if ('static' === $mode && file_exists($llms_file) && $this->delivery_probe_is_due()) { | |
| 1215 | + $this->verify_static_delivery(['message' => '']); | |
| 1216 | + $mode = $this->resolve_delivery_mode(); | |
| 1217 | + } | |
| 1218 | + | |
| 873 | 1219 | $stored = $this->get_published_content(); |
| 874 | 1220 | |
| 875 | 1221 | $status = [ |
| 876 | 1222 | 'file_exists' => file_exists($llms_file), |
| @@ -875,13 +1221,19 @@ | ||
| 875 | 1221 | $status = [ |
| 876 | 1222 | 'file_exists' => file_exists($llms_file), |
| 877 | 1223 | // Whether /llms.txt is actually being served, either mode. Prefer |
| 878 | 1224 | // this over file_exists, which is only meaningful in static mode. |
| 879 | - 'published' => file_exists($llms_file) || '' !== trim($stored), | |
| 1225 | + 'published' => $this->is_published(), | |
| 880 | 1226 | 'delivery_mode' => $mode, |
| 881 | 1227 | 'file_path' => 'dynamic' === $mode ? '' : $llms_file, |
| 882 | 1228 | 'file_url' => home_url('/llms.txt'), |
| 883 | 1229 | 'writable' => $this->is_directory_writable(dirname($llms_file)), |
| 1230 | + // Non-empty only when the site is on static delivery that the server | |
| 1231 | + // is known to answer without a charset — i.e. an explicit `static` | |
| 1232 | + // the plugin will not overrule, which is the user's to fix. | |
| 1233 | + 'delivery_warning' => 'static' === $mode && $this->static_delivery_drops_charset() | |
| 1234 | + ? self::STATIC_CHARSET_WARNING | |
| 1235 | + : '', | |
| 884 | 1236 | 'last_modified' => null, |
| 885 | 1237 | 'file_size' => null, |
| 886 | 1238 | 'content_preview' => '' |
| 887 | 1239 | ]; |
| @@ -907,12 +1259,11 @@ | ||
| 907 | 1259 | } |
| 908 | 1260 | |
| 909 | 1261 | if (null !== $content) { |
| 910 | 1262 | // Get content preview (first 200 characters) with size safety |
| 911 | - $status['content_preview'] = substr($content, 0, 200); | |
| 912 | - if (strlen($content) > 200) { | |
| 913 | - $status['content_preview'] .= '...'; | |
| 914 | - } | |
| 1263 | + // substr()/strlen() count BYTES, so this cut a multibyte character | |
| 1264 | + // in half and shipped an invalid UTF-8 sequence in the preview (#687). | |
| 1265 | + $status['content_preview'] = \ThinkRank\Core\Seo_Text::trim_to_length($content, 200); | |
| 915 | 1266 | } |
| 916 | 1267 | |
| 917 | 1268 | // Cache the result for 5 minutes to improve performance |
| 918 | 1269 | set_transient($cache_key, $status, 5 * MINUTE_IN_SECONDS); |
| @@ -1248,11 +1599,20 @@ | ||
| 1248 | 1599 | } |
| 1249 | 1600 | |
| 1250 | 1601 | // Check key features quality |
| 1251 | 1602 | if (!empty($user_input['key_features'])) { |
| 1252 | - $features = explode("\n", $user_input['key_features']); | |
| 1253 | - $feature_count = count(array_filter($features, 'trim')); | |
| 1603 | + // The same splitter the generated file uses, so the count reported | |
| 1604 | + // here and the bullets written out can never disagree (#765). | |
| 1605 | + $feature_count = count(self::split_key_features((string) $user_input['key_features'])); | |
| 1254 | 1606 | |
| 1607 | + // A single line containing commas is ambiguous: it is either a | |
| 1608 | + // legacy comma-separated list or one feature with a comma in it. | |
| 1609 | + // Rather than guess and risk publishing "and Etsy" as a feature, | |
| 1610 | + // say so and let the author decide. | |
| 1611 | + if (self::looks_like_comma_list((string) $user_input['key_features'])) { | |
| 1612 | + $validation['suggestions'][] = 'Put each key feature on its own line. Commas are treated as part of a feature, not as separators.'; | |
| 1613 | + } | |
| 1614 | + | |
| 1255 | 1615 | if ($feature_count < 3) { |
| 1256 | 1616 | $validation['warnings'][] = 'Consider adding more key features (3-8 recommended) for comprehensive AI understanding'; |
| 1257 | 1617 | $validation['score'] -= 10; |
| 1258 | 1618 | } elseif ($feature_count > 10) { |
| @@ -1713,9 +2073,9 @@ | ||
| 1713 | 2073 | ], |
| 1714 | 2074 | 'key_features' => [ |
| 1715 | 2075 | 'type' => 'string', |
| 1716 | 2076 | 'title' => 'Key Features', |
| 1717 | - 'description' => 'Main features and functionality of your website', | |
| 2077 | + 'description' => 'Main features and functionality of your website. One feature per line: a comma is treated as part of a feature, not as a separator.', | |
| 1718 | 2078 | 'default' => '', |
| 1719 | 2079 | 'maxLength' => 500 |
| 1720 | 2080 | ], |
| 1721 | 2081 | 'target_audience' => [ |
| @@ -1823,8 +2183,243 @@ | ||
| 1823 | 2183 | return "> " . $description . "\n\n"; |
| 1824 | 2184 | } |
| 1825 | 2185 | |
| 1826 | 2186 | /** |
| 2187 | + * Split the Key Features field into individual features. | |
| 2188 | + * | |
| 2189 | + * One feature per line. Commas used to be delimiters too, which meant a | |
| 2190 | + * single feature that happened to contain one — "Collect reviews from | |
| 2191 | + * Trustpilot, Google, and Etsy" — was published as three bullets, one of | |
| 2192 | + * them reading "and Etsy" (#765). Validation counted by newline only, so | |
| 2193 | + * it reported one feature while the file showed three and never flagged | |
| 2194 | + * the split. | |
| 2195 | + * | |
| 2196 | + * Commas are not a fallback delimiter even when the value has no newlines. | |
| 2197 | + * A comma inside a feature is ordinary prose and far more likely than a | |
| 2198 | + * deliberate comma-separated list, and guessing wrong publishes mangled | |
| 2199 | + * text to the file AI crawlers read. A single-line value with commas is | |
| 2200 | + * kept whole and validate_content_quality() suggests splitting it, which | |
| 2201 | + * tells the user what to do instead of quietly deciding for them. | |
| 2202 | + * | |
| 2203 | + * The one splitter both generation and validation use, so the file and the | |
| 2204 | + * feature count can no longer disagree. | |
| 2205 | + * | |
| 2206 | + * @since 2.10.0 | |
| 2207 | + * | |
| 2208 | + * @param string $key_features Raw field value. | |
| 2209 | + * @return string[] Trimmed features, empties removed. | |
| 2210 | + */ | |
| 2211 | + public static function split_key_features(string $key_features): array { | |
| 2212 | + $features = preg_split('/[\r\n]+/', $key_features); | |
| 2213 | + | |
| 2214 | + if (!is_array($features)) { | |
| 2215 | + return []; | |
| 2216 | + } | |
| 2217 | + | |
| 2218 | + $features = array_map('trim', $features); | |
| 2219 | + | |
| 2220 | + return array_values(array_filter($features, static fn(string $f): bool => '' !== $f)); | |
| 2221 | + } | |
| 2222 | + | |
| 2223 | + /** | |
| 2224 | + * Whether a value looks like the old comma-separated list. | |
| 2225 | + * | |
| 2226 | + * One line, and a comma in it. That is either a legacy list saved before | |
| 2227 | + * newlines became the delimiter, or a single feature containing a comma — | |
| 2228 | + * indistinguishable from the outside, which is exactly why this prompts | |
| 2229 | + * rather than splits. | |
| 2230 | + * | |
| 2231 | + * @since 2.10.0 | |
| 2232 | + * | |
| 2233 | + * @param string $key_features Raw field value. | |
| 2234 | + * @return bool | |
| 2235 | + */ | |
| 2236 | + public static function looks_like_comma_list(string $key_features): bool { | |
| 2237 | + $trimmed = trim($key_features); | |
| 2238 | + | |
| 2239 | + if ('' === $trimmed || false !== strpbrk($trimmed, "\r\n")) { | |
| 2240 | + return false; | |
| 2241 | + } | |
| 2242 | + | |
| 2243 | + return false !== strpos($trimmed, ','); | |
| 2244 | + } | |
| 2245 | + | |
| 2246 | + /** | |
| 2247 | + * Turn a single-line comma list into one feature per line. | |
| 2248 | + * | |
| 2249 | + * Returns null when the value is not something to convert: it already has | |
| 2250 | + * line breaks, has no comma, or reads as one feature containing a series. | |
| 2251 | + * | |
| 2252 | + * Only for text that was written under a comma rule: values saved before | |
| 2253 | + * newlines became the only delimiter (see | |
| 2254 | + * {@see self::maybe_migrate_legacy_key_features()}), and AI replies that | |
| 2255 | + * ignored the one-per-line instruction. Typed input never goes through | |
| 2256 | + * this; split_key_features() still keeps a comma inside a feature (#765). | |
| 2257 | + * | |
| 2258 | + * A series is the one shape the old rule demonstrably mangled: "Collect | |
| 2259 | + * reviews from Trustpilot, Google, and Etsy" became three bullets, the | |
| 2260 | + * last reading "and Etsy". So a value is left whole when a segment after | |
| 2261 | + * the first opens with a conjunction (the Oxford form), or when the final | |
| 2262 | + * segment carries one ("..., Google and Etsy", the form the field's own | |
| 2263 | + * placeholder uses). A plain list that happens to end "X and Y" is left | |
| 2264 | + * whole too; validation still suggests splitting it, and one intact bullet | |
| 2265 | + * is the safer wrong answer than a sentence cut into fragments. | |
| 2266 | + * | |
| 2267 | + * A comma between digits ("1,000 templates") is a thousands separator, | |
| 2268 | + * not a delimiter. | |
| 2269 | + * | |
| 2270 | + * @since 2.10.0 | |
| 2271 | + * | |
| 2272 | + * @param string $key_features Raw value. | |
| 2273 | + * @return string|null Newline-separated features, or null to leave as is. | |
| 2274 | + */ | |
| 2275 | + public static function comma_list_to_lines(string $key_features): ?string { | |
| 2276 | + if (!self::looks_like_comma_list($key_features)) { | |
| 2277 | + return null; | |
| 2278 | + } | |
| 2279 | + | |
| 2280 | + $segments = preg_split('/\s*,(?!\d)\s*/', trim($key_features)); | |
| 2281 | + | |
| 2282 | + if (!is_array($segments)) { | |
| 2283 | + return null; | |
| 2284 | + } | |
| 2285 | + | |
| 2286 | + $segments = array_values(array_filter( | |
| 2287 | + array_map('trim', $segments), | |
| 2288 | + static fn(string $s): bool => '' !== $s | |
| 2289 | + )); | |
| 2290 | + | |
| 2291 | + if (count($segments) < 2) { | |
| 2292 | + return null; | |
| 2293 | + } | |
| 2294 | + | |
| 2295 | + foreach (array_slice($segments, 1) as $segment) { | |
| 2296 | + if (preg_match('/^(?:(?:and|or|nor|plus)\b|&)/i', $segment)) { | |
| 2297 | + return null; | |
| 2298 | + } | |
| 2299 | + } | |
| 2300 | + | |
| 2301 | + if (preg_match('/\s(?:and|or|&)\s/i', (string) end($segments))) { | |
| 2302 | + return null; | |
| 2303 | + } | |
| 2304 | + | |
| 2305 | + return implode("\n", $segments); | |
| 2306 | + } | |
| 2307 | + | |
| 2308 | + /** | |
| 2309 | + * Coerce an AI reply for Key Features into the one-per-line field value. | |
| 2310 | + * | |
| 2311 | + * The prompt asks for one feature per line, but models still answer with a | |
| 2312 | + * JSON array or a comma-separated line. An array went through | |
| 2313 | + * sanitize_textarea_field() as '' and the field silently kept its old | |
| 2314 | + * value; a comma line was published as a single bullet now that commas are | |
| 2315 | + * not delimiters. Both are normalised to lines here, before sanitising. | |
| 2316 | + * | |
| 2317 | + * @since 2.10.0 | |
| 2318 | + * | |
| 2319 | + * @param mixed $value Decoded `key_features` from the reply. | |
| 2320 | + * @return string Sanitised, newline-separated features. | |
| 2321 | + */ | |
| 2322 | + public static function normalize_ai_key_features($value): string { | |
| 2323 | + if (is_array($value)) { | |
| 2324 | + $features = []; | |
| 2325 | + foreach ($value as $item) { | |
| 2326 | + if (is_scalar($item)) { | |
| 2327 | + $item = trim((string) $item); | |
| 2328 | + if ('' !== $item) { | |
| 2329 | + $features[] = $item; | |
| 2330 | + } | |
| 2331 | + } | |
| 2332 | + } | |
| 2333 | + $value = implode("\n", $features); | |
| 2334 | + } elseif (!is_scalar($value)) { | |
| 2335 | + return ''; | |
| 2336 | + } | |
| 2337 | + | |
| 2338 | + $value = (string) $value; | |
| 2339 | + $lines = self::comma_list_to_lines($value); | |
| 2340 | + | |
| 2341 | + return sanitize_textarea_field(null === $lines ? $value : $lines); | |
| 2342 | + } | |
| 2343 | + | |
| 2344 | + /** | |
| 2345 | + * Convert a Key Features value saved under the old comma rule, once. | |
| 2346 | + * | |
| 2347 | + * Up to 2.9.0 a comma separated features, so a site that saved | |
| 2348 | + * "SEO audits, Schema markup, XML sitemaps" published three bullets. After | |
| 2349 | + * #765 made newlines the only delimiter the same stored value regenerates | |
| 2350 | + * as one bullet holding the whole line, a silent change to the file AI | |
| 2351 | + * crawlers read. Rewriting the stored value as lines keeps that site's | |
| 2352 | + * output what it was, in the form the field now documents. | |
| 2353 | + * | |
| 2354 | + * A migration rather than a runtime fallback on purpose: a fallback would | |
| 2355 | + * keep treating commas as delimiters for every single-line value forever, | |
| 2356 | + * which is the #765 bug. Here only values that were saved while commas | |
| 2357 | + * really were delimiters are touched, exactly once; anything typed after | |
| 2358 | + * this has run follows the new rule. comma_list_to_lines() still leaves a | |
| 2359 | + * series such as the #765 value whole. | |
| 2360 | + * | |
| 2361 | + * Version-gated like Settings::retire_seeded_ai_provider(), and the marker | |
| 2362 | + * is written first so a site that fails the write does not retry on every | |
| 2363 | + * admin request. The activator records it on a fresh install. | |
| 2364 | + * | |
| 2365 | + * @since 2.10.0 | |
| 2366 | + * | |
| 2367 | + * @return void | |
| 2368 | + */ | |
| 2369 | + public static function maybe_migrate_legacy_key_features(): void { | |
| 2370 | + if (get_option(self::KEY_FEATURES_MIGRATION_OPTION) === self::KEY_FEATURES_MIGRATION_VERSION) { | |
| 2371 | + return; | |
| 2372 | + } | |
| 2373 | + | |
| 2374 | + update_option(self::KEY_FEATURES_MIGRATION_OPTION, self::KEY_FEATURES_MIGRATION_VERSION, true); | |
| 2375 | + | |
| 2376 | + (new static())->migrate_stored_key_features(); | |
| 2377 | + } | |
| 2378 | + | |
| 2379 | + /** | |
| 2380 | + * Rewrite the stored Key Features as lines when it is a legacy comma list. | |
| 2381 | + * | |
| 2382 | + * @since 2.10.0 | |
| 2383 | + * | |
| 2384 | + * @return bool True when a value was converted and saved. | |
| 2385 | + */ | |
| 2386 | + public function migrate_stored_key_features(): bool { | |
| 2387 | + $stored = $this->get_stored_settings('site'); | |
| 2388 | + | |
| 2389 | + if (!isset($stored['key_features']) || !is_string($stored['key_features'])) { | |
| 2390 | + return false; | |
| 2391 | + } | |
| 2392 | + | |
| 2393 | + $lines = self::comma_list_to_lines($stored['key_features']); | |
| 2394 | + | |
| 2395 | + if (null === $lines) { | |
| 2396 | + return false; | |
| 2397 | + } | |
| 2398 | + | |
| 2399 | + // validate_settings() rejects a payload without `enabled`, so carry the | |
| 2400 | + // stored flag along. Written through the base save, not this class's, | |
| 2401 | + // which would also reconcile the delivery mode: a stored-value rewrite | |
| 2402 | + // must not republish anything. | |
| 2403 | + return $this->write_migrated_key_features([ | |
| 2404 | + 'enabled' => $stored['enabled'] ?? true, | |
| 2405 | + 'key_features' => $lines, | |
| 2406 | + ]); | |
| 2407 | + } | |
| 2408 | + | |
| 2409 | + /** | |
| 2410 | + * Persist the converted value. Separate so tests can observe the write. | |
| 2411 | + * | |
| 2412 | + * @since 2.10.0 | |
| 2413 | + * | |
| 2414 | + * @param array $settings `enabled` and `key_features`. | |
| 2415 | + * @return bool | |
| 2416 | + */ | |
| 2417 | + protected function write_migrated_key_features(array $settings): bool { | |
| 2418 | + return parent::save_settings('site', null, $settings); | |
| 2419 | + } | |
| 2420 | + | |
| 2421 | + /** | |
| 1827 | 2422 | * Build additional details section |
| 1828 | 2423 | * |
| 1829 | 2424 | * @since 1.0.0 |
| 1830 | 2425 | * |
| @@ -1842,18 +2437,10 @@ | ||
| 1842 | 2437 | } |
| 1843 | 2438 | |
| 1844 | 2439 | if (!empty($key_features)) { |
| 1845 | 2440 | $content .= "**Key Features:**\n"; |
| 1846 | - // The UI field is a multi-line textarea and validation counts by | |
| 1847 | - // newline, so split on newlines (and still tolerate commas) rather | |
| 1848 | - // than commas only — otherwise newline-separated input collapses | |
| 1849 | - // into one broken bullet. | |
| 1850 | - $features = preg_split('/[\r\n,]+/', $key_features); | |
| 1851 | - foreach ($features as $feature) { | |
| 1852 | - $feature = trim($feature); | |
| 1853 | - if (!empty($feature)) { | |
| 1854 | - $content .= "- " . $feature . "\n"; | |
| 1855 | - } | |
| 2441 | + foreach (self::split_key_features($key_features) as $feature) { | |
| 2442 | + $content .= "- " . $feature . "\n"; | |
| 1856 | 2443 | } |
| 1857 | 2444 | $content .= "\n"; |
| 1858 | 2445 | } |
| 1859 | 2446 | |
| @@ -1887,9 +2474,9 @@ | ||
| 1887 | 2474 | $content .= "- [Technical Stack]({$website_url}): Built with {$stack}\n"; |
| 1888 | 2475 | } |
| 1889 | 2476 | |
| 1890 | 2477 | if (!empty($user_input['development_approach'])) { |
| 1891 | - $approach_summary = wp_trim_words($user_input['development_approach'], 10); | |
| 2478 | + $approach_summary = \ThinkRank\Core\Seo_Text::trim_words($user_input['development_approach'], 10); | |
| 1892 | 2479 | $content .= "- [Development Guidelines]({$website_url}): {$approach_summary}\n"; |
| 1893 | 2480 | } |
| 1894 | 2481 | |
| 1895 | 2482 | // Add robots.txt reference |
