| @@ -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 | } |
| @@ -51,8 +56,27 @@ | ||
| 51 | 56 | */ |
| 52 | 57 | private bool $last_unpublish_failed = false; |
| 53 | 58 | |
| 54 | 59 | /** |
| 60 | + * Message from the last save that switched delivery mode but could not move | |
| 61 | + * the published document, or an empty string when the switch was clean. | |
| 62 | + * | |
| 63 | + * @since 2.1.0 | |
| 64 | + * @var string | |
| 65 | + */ | |
| 66 | + private string $last_delivery_warning = ''; | |
| 67 | + | |
| 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 | + /** | |
| 55 | 79 | * LLMs.txt content sections configuration |
| 56 | 80 | * |
| 57 | 81 | * @since 1.0.0 |
| 58 | 82 | * @var array |
| @@ -66,9 +90,9 @@ | ||
| 66 | 90 | ], |
| 67 | 91 | 'key_features' => [ |
| 68 | 92 | 'title' => 'Key Features', |
| 69 | 93 | 'required' => true, |
| 70 | - '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.', | |
| 71 | 95 | 'max_length' => 300 |
| 72 | 96 | ], |
| 73 | 97 | 'architecture' => [ |
| 74 | 98 | 'title' => 'Architecture & Components', |
| @@ -119,8 +143,80 @@ | ||
| 119 | 143 | */ |
| 120 | 144 | private const HTACCESS_MARKER = 'ThinkRank llms.txt'; |
| 121 | 145 | |
| 122 | 146 | /** |
| 147 | + * Option holding the published llms.txt document. | |
| 148 | + * | |
| 149 | + * The published content lives here regardless of delivery mode, so the | |
| 150 | + * dynamic route has an authoritative source that does not depend on a | |
| 151 | + * physical file, and switching modes never loses the published document. | |
| 152 | + * | |
| 153 | + * @since 2.1.0 | |
| 154 | + * @var string | |
| 155 | + */ | |
| 156 | + private const CONTENT_OPTION = 'thinkrank_llms_txt_content'; | |
| 157 | + | |
| 158 | + /** | |
| 159 | + * Option holding the Unix timestamp of the last publish. | |
| 160 | + * | |
| 161 | + * @since 2.1.0 | |
| 162 | + * @var string | |
| 163 | + */ | |
| 164 | + private const PUBLISHED_AT_OPTION = 'thinkrank_llms_txt_published_at'; | |
| 165 | + | |
| 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 | + /** | |
| 199 | + * Delivery modes accepted by the `delivery_mode` setting. | |
| 200 | + * | |
| 201 | + * @since 2.1.0 | |
| 202 | + * @var string[] | |
| 203 | + */ | |
| 204 | + private const DELIVERY_MODES = ['auto', 'static', 'dynamic']; | |
| 205 | + | |
| 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 | + /** | |
| 123 | 219 | * Business type templates for content generation |
| 124 | 220 | * |
| 125 | 221 | * @since 1.0.0 |
| 126 | 222 | * @var array |
| @@ -212,8 +308,13 @@ | ||
| 212 | 308 | 'validation' => [], |
| 213 | 309 | 'file_info' => [] |
| 214 | 310 | ]; |
| 215 | 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 | + | |
| 216 | 317 | // Get current settings |
| 217 | 318 | $settings = $this->get_settings('site'); |
| 218 | 319 | |
| 219 | 320 | // Merge saved settings underneath the provided input so that empty or |
| @@ -279,10 +380,18 @@ | ||
| 279 | 380 | * @return bool |
| 280 | 381 | */ |
| 281 | 382 | public function save_settings(string $context_type, ?int $context_id, array $settings): bool { |
| 282 | 383 | $this->last_unpublish_failed = false; |
| 384 | + $this->last_delivery_warning = ''; | |
| 385 | + $this->last_delivery_switch_failed = false; | |
| 386 | + $previous_mode = $this->resolve_delivery_mode(); | |
| 387 | + | |
| 283 | 388 | $result = parent::save_settings($context_type, $context_id, $settings); |
| 284 | 389 | |
| 390 | + // The cached status carries the resolved delivery mode, so it goes stale | |
| 391 | + // the moment settings change — even when nothing needs republishing. | |
| 392 | + delete_transient('thinkrank_llms_file_status'); | |
| 393 | + | |
| 285 | 394 | // When a save explicitly disables the feature, delete the published file. |
| 286 | 395 | if ($result && array_key_exists('enabled', $settings) && empty($settings['enabled'])) { |
| 287 | 396 | if (!$this->delete_llms_txt_file()) { |
| 288 | 397 | // The settings were persisted, but the physical file could not be |
| @@ -289,14 +398,124 @@ | ||
| 289 | 398 | // removed, so /llms.txt may still be served. Record it so callers |
| 290 | 399 | // report a partial failure instead of an unqualified success. |
| 291 | 400 | $this->last_unpublish_failed = true; |
| 292 | 401 | } |
| 402 | + | |
| 403 | + return $result; | |
| 293 | 404 | } |
| 294 | 405 | |
| 406 | + // The published document has to sit where the active mode serves it from, | |
| 407 | + // or the site keeps answering on the old path: a leftover physical file | |
| 408 | + // shadows the dynamic route on every stack, and a database-only document | |
| 409 | + // is invisible to a stack now expecting a file. Reconciled on any save, | |
| 410 | + // not just an explicit mode change, so a site whose auto-detection now | |
| 411 | + // resolves differently — an nginx install upgrading into this fix with a | |
| 412 | + // static file already on disk — heals the next time settings are saved. | |
| 413 | + if ($result && $this->delivery_needs_reconcile($previous_mode)) { | |
| 414 | + $this->republish_for_delivery_mode(); | |
| 415 | + } | |
| 416 | + | |
| 295 | 417 | return $result; |
| 296 | 418 | } |
| 297 | 419 | |
| 298 | 420 | /** |
| 421 | + * Whether the published document is out of step with the active mode. | |
| 422 | + * | |
| 423 | + * @since 2.1.0 | |
| 424 | + * | |
| 425 | + * @param string $previous_mode Mode in force before the save. | |
| 426 | + * @return bool | |
| 427 | + */ | |
| 428 | + private function delivery_needs_reconcile(string $previous_mode): bool { | |
| 429 | + $mode = $this->resolve_delivery_mode(); | |
| 430 | + $file_exists = file_exists(ABSPATH . 'llms.txt'); | |
| 431 | + | |
| 432 | + if ('dynamic' === $mode) { | |
| 433 | + // A physical file would be served instead of the PHP route. | |
| 434 | + return $file_exists; | |
| 435 | + } | |
| 436 | + | |
| 437 | + // Static: a stored document with no file behind it is unreachable on a | |
| 438 | + // stack that expects one. A mode flip also forces the charset block to | |
| 439 | + // be (re)written for a file that predates it. | |
| 440 | + return (!$file_exists && '' !== trim($this->get_published_content())) | |
| 441 | + || $mode !== $previous_mode; | |
| 442 | + } | |
| 443 | + | |
| 444 | + /** | |
| 445 | + * Re-publish the current document under the active delivery mode. | |
| 446 | + * | |
| 447 | + * A no-op when nothing is published yet — this only moves an existing | |
| 448 | + * document, it never publishes on the user's behalf. | |
| 449 | + * | |
| 450 | + * @since 2.1.0 | |
| 451 | + * | |
| 452 | + * @return void | |
| 453 | + */ | |
| 454 | + private function republish_for_delivery_mode(): void { | |
| 455 | + $content = $this->get_published_content(); | |
| 456 | + | |
| 457 | + if ('' === trim($content)) { | |
| 458 | + // Published before the stored copy existed: recover it from the file. | |
| 459 | + $llms_file = ABSPATH . 'llms.txt'; | |
| 460 | + if (file_exists($llms_file)) { | |
| 461 | + $read_result = $this->safe_file_read($llms_file); | |
| 462 | + if ($read_result['success']) { | |
| 463 | + $content = $read_result['content']; | |
| 464 | + } | |
| 465 | + } | |
| 466 | + } | |
| 467 | + | |
| 468 | + if ('' === trim($content)) { | |
| 469 | + return; | |
| 470 | + } | |
| 471 | + | |
| 472 | + $write = $this->write_llms_txt_to_file($content); | |
| 473 | + | |
| 474 | + // The switch itself failed (an unwritable root on the way to static, a | |
| 475 | + // stuck file on the way to dynamic). The settings are saved, so report | |
| 476 | + // it rather than letting the mode read as applied when it is not. | |
| 477 | + if (empty($write['success'])) { | |
| 478 | + $this->last_delivery_warning = isset($write['message']) && '' !== (string) $write['message'] | |
| 479 | + ? (string) $write['message'] | |
| 480 | + : 'The delivery method was saved, but the published llms.txt could not be moved to it.'; | |
| 481 | + $this->last_delivery_switch_failed = true; | |
| 482 | + return; | |
| 483 | + } | |
| 484 | + | |
| 485 | + // The switch worked, but static delivery on this server cannot carry the | |
| 486 | + // charset the document needs. Only an explicitly chosen `static` gets | |
| 487 | + // this far — `auto` moves itself to WordPress delivery instead. | |
| 488 | + if (!empty($write['delivery_warning'])) { | |
| 489 | + $this->last_delivery_warning = (string) $write['delivery_warning']; | |
| 490 | + } | |
| 491 | + } | |
| 492 | + | |
| 493 | + /** | |
| 494 | + * Message from the last save whose delivery-mode switch could not be | |
| 495 | + * applied to the already-published document, or '' when there was none. | |
| 496 | + * | |
| 497 | + * @since 2.1.0 | |
| 498 | + * | |
| 499 | + * @return string | |
| 500 | + */ | |
| 501 | + public function delivery_switch_warning(): string { | |
| 502 | + return $this->last_delivery_warning; | |
| 503 | + } | |
| 504 | + | |
| 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 | + /** | |
| 299 | 518 | * Whether the last save_settings() disabled the feature but could not remove |
| 300 | 519 | * the published llms.txt file (which may therefore still be served). |
| 301 | 520 | * |
| 302 | 521 | * @return bool |
| @@ -305,15 +524,370 @@ | ||
| 305 | 524 | return $this->last_unpublish_failed; |
| 306 | 525 | } |
| 307 | 526 | |
| 308 | 527 | /** |
| 309 | - * Remove the published llms.txt file and bust the status transient. | |
| 528 | + * Resolve the effective delivery mode for /llms.txt. | |
| 310 | 529 | * |
| 311 | - * @return bool True if the file is absent or was removed. | |
| 530 | + * `static` publishes a physical ABSPATH/llms.txt and lets the web server | |
| 531 | + * answer it; `dynamic` keeps the document in the database and lets the PHP | |
| 532 | + * route in {@see serve_llms_txt()} answer it. `auto` picks static only on | |
| 533 | + * Apache/LiteSpeed, the stacks that read the .htaccess charset block — on | |
| 534 | + * nginx a physical file is served with a bare `Content-Type: text/plain` | |
| 535 | + * that neither fix path can reach, which renders UTF-8 as mojibake (#419). | |
| 536 | + * | |
| 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. | |
| 545 | + * | |
| 546 | + * @since 2.1.0 | |
| 547 | + * | |
| 548 | + * @param string|null $mode Optional. Raw setting value; read from the saved | |
| 549 | + * settings when null. | |
| 550 | + * @return string Either 'static' or 'dynamic'. | |
| 312 | 551 | */ |
| 552 | + public function resolve_delivery_mode(?string $mode = null): string { | |
| 553 | + if (null === $mode) { | |
| 554 | + $settings = $this->get_settings('site'); | |
| 555 | + $mode = (string) ($settings['delivery_mode'] ?? 'auto'); | |
| 556 | + } | |
| 557 | + | |
| 558 | + if ('static' === $mode || 'dynamic' === $mode) { | |
| 559 | + return $mode; | |
| 560 | + } | |
| 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 | + | |
| 572 | + // $is_apache also covers LiteSpeed, which reads .htaccess the same way. | |
| 573 | + if (empty($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 $this->static_delivery_drops_charset() ? 'dynamic' : 'static'; | |
| 581 | + } | |
| 582 | + | |
| 583 | + /** | |
| 584 | + * Whether the recorded check caught the public URL dropping the charset. | |
| 585 | + * | |
| 586 | + * @since 2.1.0 | |
| 587 | + * | |
| 588 | + * @return bool | |
| 589 | + */ | |
| 590 | + private function static_delivery_drops_charset(): bool { | |
| 591 | + return 'no_charset' === ($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 | + /** | |
| 768 | + * The published llms.txt document, or an empty string when unpublished. | |
| 769 | + * | |
| 770 | + * @since 2.1.0 | |
| 771 | + * | |
| 772 | + * @return string | |
| 773 | + */ | |
| 774 | + public function get_published_content(): string { | |
| 775 | + $content = get_option(self::CONTENT_OPTION, ''); | |
| 776 | + | |
| 777 | + return is_string($content) ? $content : ''; | |
| 778 | + } | |
| 779 | + | |
| 780 | + /** | |
| 781 | + * Whether /llms.txt is currently being served, in either delivery mode. | |
| 782 | + * | |
| 783 | + * `static` publishes a file at ABSPATH; `dynamic` keeps the document in | |
| 784 | + * an option and answers from serve_llms_txt(). Callers that only need | |
| 785 | + * this yes/no must use it in preference to get_llms_txt_status(), which | |
| 786 | + * resolves the delivery mode, may fire a loopback delivery probe, asks | |
| 787 | + * the filesystem API whether ABSPATH is writable, reads the document and | |
| 788 | + * writes a transient — far too much work for a boolean, and not | |
| 789 | + * something a dashboard summary should be triggering. | |
| 790 | + * | |
| 791 | + * @since 2.2.1 | |
| 792 | + * | |
| 793 | + * @return bool | |
| 794 | + */ | |
| 795 | + public function is_published(): bool { | |
| 796 | + return file_exists(ABSPATH . 'llms.txt') | |
| 797 | + || '' !== trim($this->get_published_content()); | |
| 798 | + } | |
| 799 | + | |
| 800 | + /** | |
| 801 | + * Ask the common page/CDN cache layers to drop their copy of /llms.txt. | |
| 802 | + * | |
| 803 | + * A cached response outlives a republish, so without this a mode switch or | |
| 804 | + * a content change keeps serving the old document (and, on the static path, | |
| 805 | + * the old headers). Every call is guarded — a site running none of these | |
| 806 | + * simply gets the action hook, which integrations can use. | |
| 807 | + * | |
| 808 | + * @since 2.1.0 | |
| 809 | + * | |
| 810 | + * @return void | |
| 811 | + */ | |
| 812 | + private function purge_llms_txt_caches(): void { | |
| 813 | + $url = home_url('/llms.txt'); | |
| 814 | + | |
| 815 | + /** | |
| 816 | + * Fires after the published llms.txt changes, so cache layers ThinkRank | |
| 817 | + * does not know about can drop their copy. | |
| 818 | + * | |
| 819 | + * @since 2.1.0 | |
| 820 | + * | |
| 821 | + * @param string $url Public URL of the llms.txt document. | |
| 822 | + */ | |
| 823 | + do_action('thinkrank_llms_txt_updated', $url); | |
| 824 | + | |
| 825 | + // LiteSpeed Cache and Nginx Helper both listen on their own actions. | |
| 826 | + // These are third-party hook names we fire, not ours to prefix. | |
| 827 | + do_action('litespeed_purge_url', $url); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound | |
| 828 | + do_action('rt_nginx_helper_purge_all'); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound | |
| 829 | + | |
| 830 | + if (function_exists('rocket_clean_files')) { | |
| 831 | + rocket_clean_files([$url]); | |
| 832 | + } | |
| 833 | + if (function_exists('w3tc_flush_url')) { | |
| 834 | + w3tc_flush_url($url); | |
| 835 | + } | |
| 836 | + if (function_exists('wpsc_delete_url_cache')) { | |
| 837 | + wpsc_delete_url_cache($url); | |
| 838 | + } | |
| 839 | + } | |
| 840 | + | |
| 841 | + /** | |
| 842 | + * Unpublish llms.txt: drop the stored document and any physical file. | |
| 843 | + * | |
| 844 | + * Both delivery modes are cleared, not just the active one, so a site that | |
| 845 | + * published under one mode and switched to the other is left with nothing | |
| 846 | + * still being served. | |
| 847 | + * | |
| 848 | + * @return bool True once nothing is left to serve. | |
| 849 | + */ | |
| 313 | 850 | public function delete_llms_txt_file(): bool { |
| 851 | + delete_option(self::CONTENT_OPTION); | |
| 852 | + delete_option(self::PUBLISHED_AT_OPTION); | |
| 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 { | |
| 314 | 874 | delete_transient('thinkrank_llms_file_status'); |
| 315 | 875 | |
| 876 | + $removed = $this->delete_static_file(); | |
| 877 | + $this->purge_llms_txt_caches(); | |
| 878 | + | |
| 879 | + return $removed; | |
| 880 | + } | |
| 881 | + | |
| 882 | + /** | |
| 883 | + * Remove the physical ABSPATH/llms.txt and its .htaccess charset block. | |
| 884 | + * | |
| 885 | + * @since 2.1.0 | |
| 886 | + * | |
| 887 | + * @return bool True if the file is absent or was removed. | |
| 888 | + */ | |
| 889 | + private function delete_static_file(): bool { | |
| 316 | 890 | $llms_file = ABSPATH . 'llms.txt'; |
| 317 | 891 | if (!file_exists($llms_file)) { |
| 318 | 892 | $this->remove_htaccess_charset(); |
| 319 | 893 | return true; |
| @@ -445,9 +1019,12 @@ | ||
| 445 | 1019 | |
| 446 | 1020 | $content = ''; |
| 447 | 1021 | $llms_file = ABSPATH . 'llms.txt'; |
| 448 | 1022 | |
| 449 | - if (file_exists($llms_file)) { | |
| 1023 | + // In static mode a physical file is what the server would normally hand | |
| 1024 | + // back, so prefer its exact bytes; in dynamic mode there is no file and | |
| 1025 | + // the stored document is the authoritative copy. | |
| 1026 | + if ('static' === $this->resolve_delivery_mode() && file_exists($llms_file)) { | |
| 450 | 1027 | $read_result = $this->safe_file_read($llms_file); |
| 451 | 1028 | if ($read_result['success']) { |
| 452 | 1029 | $content = $read_result['content']; |
| 453 | 1030 | } |
| @@ -453,8 +1030,12 @@ | ||
| 453 | 1030 | } |
| 454 | 1031 | } |
| 455 | 1032 | |
| 456 | 1033 | if ('' === trim($content)) { |
| 1034 | + $content = $this->get_published_content(); | |
| 1035 | + } | |
| 1036 | + | |
| 1037 | + if ('' === trim($content)) { | |
| 457 | 1038 | $generated = $this->generate_llms_txt([]); |
| 458 | 1039 | $content = (string) ($generated['content'] ?? ''); |
| 459 | 1040 | } |
| 460 | 1041 | |
| @@ -496,10 +1077,39 @@ | ||
| 496 | 1077 | $result['message'] = 'LLMs.txt is disabled. Enable it before publishing.'; |
| 497 | 1078 | return $result; |
| 498 | 1079 | } |
| 499 | 1080 | |
| 1081 | + $mode = $this->resolve_delivery_mode(); | |
| 1082 | + $result['delivery_mode'] = $mode; | |
| 1083 | + | |
| 500 | 1084 | $llms_file = ABSPATH . 'llms.txt'; |
| 1085 | + $result['file_path'] = $llms_file; | |
| 501 | 1086 | |
| 1087 | + // Dynamic delivery: the document lives in the database and /llms.txt is | |
| 1088 | + // answered by serve_llms_txt(), which sets `charset=utf-8` itself. A | |
| 1089 | + // physical file would shadow that route on every stack, so any leftover | |
| 1090 | + // from a previous static publish has to go. | |
| 1091 | + if ('dynamic' === $mode) { | |
| 1092 | + if (!$this->delete_static_file()) { | |
| 1093 | + $result['message'] = 'A physical llms.txt is still present and could not be removed. It would be served instead of the dynamic route.'; | |
| 1094 | + return $result; | |
| 1095 | + } | |
| 1096 | + | |
| 1097 | + $this->store_published_content($content); | |
| 1098 | + | |
| 1099 | + $result['success'] = true; | |
| 1100 | + $result['message'] = 'LLMs.txt published. It is served by WordPress as UTF-8 text.'; | |
| 1101 | + $result['bytes_written'] = strlen($content); | |
| 1102 | + $result['charset_pinned'] = true; | |
| 1103 | + $result['permissions'] = [ | |
| 1104 | + 'directory_writable' => $this->is_directory_writable(ABSPATH), | |
| 1105 | + 'file_exists' => false, | |
| 1106 | + 'file_writable' => null, | |
| 1107 | + ]; | |
| 1108 | + | |
| 1109 | + return $result; | |
| 1110 | + } | |
| 1111 | + | |
| 502 | 1112 | // Security: Validate file path to prevent path traversal attacks |
| 503 | 1113 | $real_llms_file = realpath(dirname($llms_file)) . DIRECTORY_SEPARATOR . basename($llms_file); |
| 504 | 1114 | $allowed_dir = realpath(ABSPATH); |
| 505 | 1115 | |
| @@ -507,10 +1117,8 @@ | ||
| 507 | 1117 | $result['message'] = 'Invalid file path detected for security reasons.'; |
| 508 | 1118 | return $result; |
| 509 | 1119 | } |
| 510 | 1120 | |
| 511 | - $result['file_path'] = $llms_file; | |
| 512 | - | |
| 513 | 1121 | // Check directory permissions |
| 514 | 1122 | $result['permissions'] = [ |
| 515 | 1123 | 'directory_writable' => $this->is_directory_writable(ABSPATH), |
| 516 | 1124 | 'file_exists' => file_exists($llms_file), |
| @@ -528,36 +1136,57 @@ | ||
| 528 | 1136 | $result['message'] = 'Existing llms.txt file is not writable. Please check file permissions.'; |
| 529 | 1137 | return $result; |
| 530 | 1138 | } |
| 531 | 1139 | |
| 532 | - // Write new content using WP_Filesystem | |
| 533 | - if (!$this->init_filesystem()) { | |
| 534 | - $result['message'] = 'Could not initialize WordPress filesystem.'; | |
| 535 | - return $result; | |
| 536 | - } | |
| 1140 | + // Write new content using WP_Filesystem | |
| 1141 | + if (!$this->init_filesystem()) { | |
| 1142 | + $result['message'] = 'Could not initialize WordPress filesystem.'; | |
| 1143 | + return $result; | |
| 1144 | + } | |
| 537 | 1145 | |
| 538 | - $write_success = $this->filesystem->put_contents($llms_file, $content, FS_CHMOD_FILE); | |
| 1146 | + if (!$this->filesystem->put_contents($llms_file, $content, FS_CHMOD_FILE)) { | |
| 1147 | + $result['message'] = 'Failed to write llms.txt file.'; | |
| 1148 | + return $result; | |
| 1149 | + } | |
| 539 | 1150 | |
| 540 | - if ($write_success) { | |
| 541 | - $result['success'] = true; | |
| 542 | - $result['message'] = 'LLMs.txt file written successfully.'; | |
| 543 | - $result['bytes_written'] = strlen($content); | |
| 1151 | + $result['success'] = true; | |
| 1152 | + $result['message'] = 'LLMs.txt file written successfully.'; | |
| 1153 | + $result['bytes_written'] = strlen($content); | |
| 544 | 1154 | |
| 545 | - // Pin the served charset to UTF-8. Best effort: a site without a | |
| 546 | - // writable .htaccess (or not on Apache/LiteSpeed) still gets a | |
| 547 | - // correctly written file, so this must never fail the publish. | |
| 548 | - $result['charset_pinned'] = $this->sync_htaccess_charset(); | |
| 1155 | + // Pin the served charset to UTF-8. Best effort: a site without a | |
| 1156 | + // writable .htaccess (or not on Apache/LiteSpeed) still gets a | |
| 1157 | + // correctly written file, so this must never fail the publish. | |
| 1158 | + $result['charset_pinned'] = $this->sync_htaccess_charset(); | |
| 549 | 1159 | |
| 550 | - // Invalidate file status cache since file has changed | |
| 551 | - delete_transient('thinkrank_llms_file_status'); | |
| 552 | - } else { | |
| 553 | - $result['message'] = 'Failed to write llms.txt file.'; | |
| 554 | - } | |
| 1160 | + // Keep the stored copy in step with the file so a later switch to | |
| 1161 | + // dynamic delivery serves the same document. | |
| 1162 | + $this->store_published_content($content); | |
| 555 | 1163 | |
| 556 | - 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); | |
| 557 | 1168 | } |
| 558 | 1169 | |
| 559 | 1170 | /** |
| 1171 | + * Persist the published document and bust the caches that mirror it. | |
| 1172 | + * | |
| 1173 | + * @since 2.1.0 | |
| 1174 | + * | |
| 1175 | + * @param string $content Published llms.txt content. | |
| 1176 | + * @return void | |
| 1177 | + */ | |
| 1178 | + private function store_published_content(string $content): void { | |
| 1179 | + update_option(self::CONTENT_OPTION, $content, false); | |
| 1180 | + update_option(self::PUBLISHED_AT_OPTION, time(), false); | |
| 1181 | + | |
| 1182 | + // Invalidate file status cache since the published document has changed | |
| 1183 | + delete_transient('thinkrank_llms_file_status'); | |
| 1184 | + | |
| 1185 | + $this->purge_llms_txt_caches(); | |
| 1186 | + } | |
| 1187 | + | |
| 1188 | + /** | |
| 560 | 1189 | * Get LLMs.txt file status and information |
| 561 | 1190 | * |
| 562 | 1191 | * @since 1.0.0 |
| 563 | 1192 | * |
| @@ -574,36 +1203,69 @@ | ||
| 574 | 1203 | } |
| 575 | 1204 | } |
| 576 | 1205 | |
| 577 | 1206 | $llms_file = ABSPATH . 'llms.txt'; |
| 1207 | + $mode = $this->resolve_delivery_mode(); | |
| 578 | 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 | + | |
| 1219 | + $stored = $this->get_published_content(); | |
| 1220 | + | |
| 579 | 1221 | $status = [ |
| 580 | 1222 | 'file_exists' => file_exists($llms_file), |
| 581 | - 'file_path' => $llms_file, | |
| 1223 | + // Whether /llms.txt is actually being served, either mode. Prefer | |
| 1224 | + // this over file_exists, which is only meaningful in static mode. | |
| 1225 | + 'published' => $this->is_published(), | |
| 1226 | + 'delivery_mode' => $mode, | |
| 1227 | + 'file_path' => 'dynamic' === $mode ? '' : $llms_file, | |
| 582 | 1228 | 'file_url' => home_url('/llms.txt'), |
| 583 | 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 | + : '', | |
| 584 | 1236 | 'last_modified' => null, |
| 585 | 1237 | 'file_size' => null, |
| 586 | 1238 | 'content_preview' => '' |
| 587 | 1239 | ]; |
| 588 | 1240 | |
| 1241 | + $content = null; | |
| 1242 | + | |
| 589 | 1243 | if ($status['file_exists']) { |
| 590 | 1244 | $status['last_modified'] = filemtime($llms_file); |
| 591 | 1245 | $status['file_size'] = filesize($llms_file); |
| 592 | 1246 | |
| 593 | - // Get content preview (first 200 characters) with size safety | |
| 594 | 1247 | $read_result = $this->safe_file_read($llms_file); |
| 595 | 1248 | if ($read_result['success']) { |
| 596 | - $status['content_preview'] = substr($read_result['content'], 0, 200); | |
| 597 | - if (strlen($read_result['content']) > 200) { | |
| 598 | - $status['content_preview'] .= '...'; | |
| 599 | - } | |
| 1249 | + $content = $read_result['content']; | |
| 600 | 1250 | } else { |
| 601 | 1251 | $status['content_preview'] = 'Error: ' . $read_result['error']; |
| 602 | 1252 | $status['read_error'] = $read_result['error']; |
| 603 | 1253 | } |
| 1254 | + } elseif ('' !== trim($stored)) { | |
| 1255 | + $published_at = (int) get_option(self::PUBLISHED_AT_OPTION, 0); | |
| 1256 | + $status['last_modified'] = $published_at > 0 ? $published_at : null; | |
| 1257 | + $status['file_size'] = strlen($stored); | |
| 1258 | + $content = $stored; | |
| 604 | 1259 | } |
| 605 | 1260 | |
| 1261 | + if (null !== $content) { | |
| 1262 | + // Get content preview (first 200 characters) with size safety | |
| 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); | |
| 1266 | + } | |
| 1267 | + | |
| 606 | 1268 | // Cache the result for 5 minutes to improve performance |
| 607 | 1269 | set_transient($cache_key, $status, 5 * MINUTE_IN_SECONDS); |
| 608 | 1270 | |
| 609 | 1271 | return $status; |
| @@ -748,13 +1410,15 @@ | ||
| 748 | 1410 | * Override parent sanitize_settings to preserve line breaks in link fields |
| 749 | 1411 | * |
| 750 | 1412 | * @since 1.0.0 |
| 751 | 1413 | * |
| 752 | - * @param array $settings Settings to sanitize | |
| 1414 | + * @param array $settings Settings to sanitize | |
| 1415 | + * @param string $context_type Context the save is for. | |
| 753 | 1416 | * @return array Sanitized settings |
| 754 | 1417 | */ |
| 755 | - protected function sanitize_settings(array $settings): array { | |
| 1418 | + protected function sanitize_settings(array $settings, string $context_type = 'site'): array { | |
| 756 | 1419 | $sanitized = []; |
| 1420 | + $known = $this->get_known_setting_keys($context_type); | |
| 757 | 1421 | |
| 758 | 1422 | // Fields that should preserve line breaks |
| 759 | 1423 | $preserve_linebreaks = [ |
| 760 | 1424 | 'documentation_links', |
| @@ -771,8 +1435,27 @@ | ||
| 771 | 1435 | |
| 772 | 1436 | foreach ($settings as $key => $value) { |
| 773 | 1437 | $sanitized_key = sanitize_key($key); |
| 774 | 1438 | |
| 1439 | + // Never store the REST envelope back as settings (see | |
| 1440 | + // Abstract_Seo_Manager::RESERVED_ENVELOPE_KEYS). | |
| 1441 | + if (in_array($sanitized_key, self::RESERVED_ENVELOPE_KEYS, true)) { | |
| 1442 | + continue; | |
| 1443 | + } | |
| 1444 | + | |
| 1445 | + // And nothing this manager does not declare (#452). | |
| 1446 | + if (!$this->is_known_setting_key($sanitized_key, $known)) { | |
| 1447 | + continue; | |
| 1448 | + } | |
| 1449 | + | |
| 1450 | + // Constrain the delivery mode to the known enum so an unexpected | |
| 1451 | + // value falls back to auto-detection rather than being stored. | |
| 1452 | + if ('delivery_mode' === $sanitized_key) { | |
| 1453 | + $mode = is_string($value) ? sanitize_key($value) : ''; | |
| 1454 | + $sanitized[$sanitized_key] = in_array($mode, self::DELIVERY_MODES, true) ? $mode : 'auto'; | |
| 1455 | + continue; | |
| 1456 | + } | |
| 1457 | + | |
| 775 | 1458 | if (is_string($value)) { |
| 776 | 1459 | if (in_array($key, $preserve_linebreaks, true)) { |
| 777 | 1460 | // Use our custom sanitization that preserves line breaks |
| 778 | 1461 | if (in_array($key, ['documentation_links', 'technical_links', 'optional_links', 'custom_sections'], true)) { |
| @@ -916,11 +1599,20 @@ | ||
| 916 | 1599 | } |
| 917 | 1600 | |
| 918 | 1601 | // Check key features quality |
| 919 | 1602 | if (!empty($user_input['key_features'])) { |
| 920 | - $features = explode("\n", $user_input['key_features']); | |
| 921 | - $feature_count = count(array_filter($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'])); | |
| 922 | 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 | + $validation['suggestions'][] = 'Put each key feature on its own line. Commas are treated as part of a feature, not as separators.'; | |
| 1613 | + } | |
| 1614 | + | |
| 923 | 1615 | if ($feature_count < 3) { |
| 924 | 1616 | $validation['warnings'][] = 'Consider adding more key features (3-8 recommended) for comprehensive AI understanding'; |
| 925 | 1617 | $validation['score'] -= 10; |
| 926 | 1618 | } elseif ($feature_count > 10) { |
| @@ -1236,12 +1928,16 @@ | ||
| 1236 | 1928 | $validation['score'] -= 5; |
| 1237 | 1929 | } |
| 1238 | 1930 | } |
| 1239 | 1931 | |
| 1240 | - // Check file permissions if enabled | |
| 1241 | - if (!empty($settings['enabled'])) { | |
| 1932 | + // Check file permissions if enabled. Only the static delivery mode needs | |
| 1933 | + // a writable root — dynamic delivery keeps the document in the database. | |
| 1934 | + $mode = $this->resolve_delivery_mode( | |
| 1935 | + isset($settings['delivery_mode']) ? (string) $settings['delivery_mode'] : null | |
| 1936 | + ); | |
| 1937 | + if (!empty($settings['enabled']) && 'static' === $mode) { | |
| 1242 | 1938 | if (!$this->is_directory_writable(ABSPATH)) { |
| 1243 | - $validation['warnings'][] = 'WordPress root directory is not writable, llms.txt cannot be automatically managed'; | |
| 1939 | + $validation['warnings'][] = 'WordPress root directory is not writable, llms.txt cannot be automatically managed. Switch delivery to "Served by WordPress" to publish without writing a file.'; | |
| 1244 | 1940 | $validation['score'] -= 10; |
| 1245 | 1941 | } |
| 1246 | 1942 | } |
| 1247 | 1943 | |
| @@ -1276,9 +1972,10 @@ | ||
| 1276 | 1972 | |
| 1277 | 1973 | // Get file status |
| 1278 | 1974 | $output['file_status'] = $this->get_llms_txt_status(); |
| 1279 | 1975 | |
| 1280 | - // If file exists, get current content safely | |
| 1976 | + // If a file is published, get its current content safely; otherwise fall | |
| 1977 | + // back to the stored document that dynamic delivery serves. | |
| 1281 | 1978 | if ($output['file_status']['file_exists']) { |
| 1282 | 1979 | $llms_file = ABSPATH . 'llms.txt'; |
| 1283 | 1980 | $read_result = $this->safe_file_read($llms_file); |
| 1284 | 1981 | if ($read_result['success']) { |
| @@ -1286,8 +1983,10 @@ | ||
| 1286 | 1983 | } else { |
| 1287 | 1984 | $output['llms_txt_content'] = ''; |
| 1288 | 1985 | $output['file_read_error'] = $read_result['error']; |
| 1289 | 1986 | } |
| 1987 | + } else { | |
| 1988 | + $output['llms_txt_content'] = $this->get_published_content(); | |
| 1290 | 1989 | } |
| 1291 | 1990 | |
| 1292 | 1991 | // Add metadata |
| 1293 | 1992 | $output['metadata'] = [ |
| @@ -1319,8 +2018,9 @@ | ||
| 1319 | 2018 | 'development_approach' => '', |
| 1320 | 2019 | 'setup_instructions' => '', |
| 1321 | 2020 | 'ai_context_custom' => '', |
| 1322 | 2021 | 'auto_generate' => false, |
| 2022 | + 'delivery_mode' => 'auto', | |
| 1323 | 2023 | 'last_generated' => null, |
| 1324 | 2024 | // Structured sections for llms.txt spec compliance |
| 1325 | 2025 | 'documentation_links' => '', |
| 1326 | 2026 | 'technical_links' => '', |
| @@ -1373,9 +2073,9 @@ | ||
| 1373 | 2073 | ], |
| 1374 | 2074 | 'key_features' => [ |
| 1375 | 2075 | 'type' => 'string', |
| 1376 | 2076 | 'title' => 'Key Features', |
| 1377 | - '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.', | |
| 1378 | 2078 | 'default' => '', |
| 1379 | 2079 | 'maxLength' => 500 |
| 1380 | 2080 | ], |
| 1381 | 2081 | 'target_audience' => [ |
| @@ -1425,8 +2125,15 @@ | ||
| 1425 | 2125 | 'title' => 'Auto-generate', |
| 1426 | 2126 | 'description' => 'Automatically regenerate llms.txt when settings change', |
| 1427 | 2127 | 'default' => false |
| 1428 | 2128 | ], |
| 2129 | + 'delivery_mode' => [ | |
| 2130 | + 'type' => 'string', | |
| 2131 | + 'title' => 'Delivery Method', | |
| 2132 | + 'description' => 'How /llms.txt is served: "static" writes a physical file the web server answers, "dynamic" keeps the document in WordPress and serves it from PHP as UTF-8, "auto" picks static on Apache/LiteSpeed and dynamic elsewhere.', | |
| 2133 | + 'enum' => self::DELIVERY_MODES, | |
| 2134 | + 'default' => 'auto' | |
| 2135 | + ], | |
| 1429 | 2136 | 'last_generated' => [ |
| 1430 | 2137 | 'type' => 'string', |
| 1431 | 2138 | 'title' => 'Last Generated', |
| 1432 | 2139 | 'description' => 'Timestamp of last generation', |
| @@ -1476,8 +2183,243 @@ | ||
| 1476 | 2183 | return "> " . $description . "\n\n"; |
| 1477 | 2184 | } |
| 1478 | 2185 | |
| 1479 | 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, $settings); | |
| 2419 | + } | |
| 2420 | + | |
| 2421 | + /** | |
| 1480 | 2422 | * Build additional details section |
| 1481 | 2423 | * |
| 1482 | 2424 | * @since 1.0.0 |
| 1483 | 2425 | * |
| @@ -1495,18 +2437,10 @@ | ||
| 1495 | 2437 | } |
| 1496 | 2438 | |
| 1497 | 2439 | if (!empty($key_features)) { |
| 1498 | 2440 | $content .= "**Key Features:**\n"; |
| 1499 | - // The UI field is a multi-line textarea and validation counts by | |
| 1500 | - // newline, so split on newlines (and still tolerate commas) rather | |
| 1501 | - // than commas only — otherwise newline-separated input collapses | |
| 1502 | - // into one broken bullet. | |
| 1503 | - $features = preg_split('/[\r\n,]+/', $key_features); | |
| 1504 | - foreach ($features as $feature) { | |
| 1505 | - $feature = trim($feature); | |
| 1506 | - if (!empty($feature)) { | |
| 1507 | - $content .= "- " . $feature . "\n"; | |
| 1508 | - } | |
| 2441 | + foreach (self::split_key_features($key_features) as $feature) { | |
| 2442 | + $content .= "- " . $feature . "\n"; | |
| 1509 | 2443 | } |
| 1510 | 2444 | $content .= "\n"; |
| 1511 | 2445 | } |
| 1512 | 2446 | |
| @@ -1540,9 +2474,9 @@ | ||
| 1540 | 2474 | $content .= "- [Technical Stack]({$website_url}): Built with {$stack}\n"; |
| 1541 | 2475 | } |
| 1542 | 2476 | |
| 1543 | 2477 | if (!empty($user_input['development_approach'])) { |
| 1544 | - $approach_summary = wp_trim_words($user_input['development_approach'], 10); | |
| 2478 | + $approach_summary = \ThinkRank\Core\Seo_Text::trim_words($user_input['development_approach'], 10); | |
| 1545 | 2479 | $content .= "- [Development Guidelines]({$website_url}): {$approach_summary}\n"; |
| 1546 | 2480 | } |
| 1547 | 2481 | |
| 1548 | 2482 | // Add robots.txt reference |
