| 1 |
<?php |
| 2 |
// If this file is called directly, abort. |
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* LLMs.txt generator. |
| 9 |
* |
| 10 |
* Serves virtual /llms.txt and /llms-full.txt routes via template_redirect, |
| 11 |
* caches output in transients, and invalidates the cache on post or |
| 12 |
* settings changes. Modeled after Metasync_Sitemap_Generator. |
| 13 |
* |
| 14 |
* @package Metasync |
| 15 |
* @subpackage Metasync/llms-txt |
| 16 |
*/ |
| 17 |
class Metasync_Llms_Txt_Generator |
| 18 |
{ |
| 19 |
const OPTION_KEY = 'metasync_llms_txt_settings'; |
| 20 |
const TRANSIENT_SHORT = 'metasync_llms_txt'; |
| 21 |
const TRANSIENT_FULL = 'metasync_llms_full_txt'; |
| 22 |
const CONFLICT_TRANSIENT = 'metasync_llms_conflict'; |
| 23 |
const CACHE_TTL = 12 * HOUR_IN_SECONDS; |
| 24 |
|
| 25 |
/** |
| 26 |
* Initialise the generator and register hooks. |
| 27 |
*/ |
| 28 |
public function __construct() |
| 29 |
{ |
| 30 |
add_action('template_redirect', array($this, 'serve_virtual_llms_txt'), 1); |
| 31 |
|
| 32 |
// Invalidate the cache whenever a post is saved, deleted, or trashed. |
| 33 |
// save_post is gated by post_type so unrelated CPTs do not bust the cache. |
| 34 |
// deleted_post / trashed_post stay unconditional — post_type cannot always |
| 35 |
// be reliably resolved after deletion. |
| 36 |
add_action('save_post', array($this, 'maybe_invalidate_cache')); |
| 37 |
add_action('deleted_post', array($this, 'invalidate_cache')); |
| 38 |
add_action('trashed_post', array($this, 'invalidate_cache')); |
| 39 |
|
| 40 |
// Invalidate when the settings are updated. |
| 41 |
add_action('update_option_' . self::OPTION_KEY, array($this, 'invalidate_cache')); |
| 42 |
add_action('add_option_' . self::OPTION_KEY, array($this, 'invalidate_cache')); |
| 43 |
|
| 44 |
// Sync robots.txt when the enabled setting changes. |
| 45 |
add_action('update_option_' . self::OPTION_KEY, array($this, 'on_settings_changed'), 10, 2); |
| 46 |
add_action('add_option_' . self::OPTION_KEY, array($this, 'on_settings_changed'), 10, 2); |
| 47 |
} |
| 48 |
|
| 49 |
/** |
| 50 |
* Serve the virtual /llms.txt or /llms-full.txt route. |
| 51 |
*/ |
| 52 |
public function serve_virtual_llms_txt() |
| 53 |
{ |
| 54 |
if (is_admin()) { |
| 55 |
return; |
| 56 |
} |
| 57 |
|
| 58 |
$request_uri = isset($_SERVER['REQUEST_URI']) ? esc_url_raw(wp_unslash($_SERVER['REQUEST_URI'])) : ''; |
| 59 |
$request_uri = strtok($request_uri, '?'); |
| 60 |
$path = rtrim(parse_url($request_uri, PHP_URL_PATH), '/'); |
| 61 |
|
| 62 |
$is_short = ($path === '/llms.txt'); |
| 63 |
$is_full = ($path === '/llms-full.txt'); |
| 64 |
|
| 65 |
if (!$is_short && !$is_full) { |
| 66 |
return; |
| 67 |
} |
| 68 |
|
| 69 |
// Let a physical file win if one exists. |
| 70 |
$filename = $is_full ? 'llms-full.txt' : 'llms.txt'; |
| 71 |
if (file_exists(ABSPATH . $filename)) { |
| 72 |
return; |
| 73 |
} |
| 74 |
|
| 75 |
$settings = $this->get_settings(); |
| 76 |
|
| 77 |
if (empty($settings['enabled'])) { |
| 78 |
return; |
| 79 |
} |
| 80 |
|
| 81 |
// Detect conflicts for admin notice purposes only — MetaSync always serves when enabled. |
| 82 |
$this->detect_plugin_conflict(); |
| 83 |
|
| 84 |
if ($is_full && empty($settings['llms_full_enabled'])) { |
| 85 |
status_header(404); |
| 86 |
return; |
| 87 |
} |
| 88 |
|
| 89 |
$transient_key = $is_full ? self::TRANSIENT_FULL : self::TRANSIENT_SHORT; |
| 90 |
$content = get_transient($transient_key); |
| 91 |
|
| 92 |
if (false === $content || $content === '') { |
| 93 |
$content = $is_full ? $this->generate_full() : $this->generate(); |
| 94 |
if ($content !== '') { |
| 95 |
set_transient($transient_key, $content, self::CACHE_TTL); |
| 96 |
} |
| 97 |
} |
| 98 |
|
| 99 |
status_header(200); |
| 100 |
header('Content-Type: text/plain; charset=utf-8'); |
| 101 |
header('X-Robots-Tag: noindex'); |
| 102 |
echo $content; |
| 103 |
exit; |
| 104 |
} |
| 105 |
|
| 106 |
/** |
| 107 |
* Build /llms.txt content. |
| 108 |
* |
| 109 |
* @return string |
| 110 |
*/ |
| 111 |
public function generate() |
| 112 |
{ |
| 113 |
$settings = $this->get_settings(); |
| 114 |
$site_name = get_bloginfo('name'); |
| 115 |
$tagline = !empty($settings['custom_description']) |
| 116 |
? $settings['custom_description'] |
| 117 |
: get_bloginfo('description'); |
| 118 |
|
| 119 |
$content = '# ' . $site_name . "\n\n"; |
| 120 |
if (!empty($tagline)) { |
| 121 |
$content .= '> ' . $tagline . "\n\n"; |
| 122 |
} |
| 123 |
|
| 124 |
$posts = $this->query_posts($settings); |
| 125 |
|
| 126 |
// Group posts by post type for cleaner section headers. |
| 127 |
$grouped = []; |
| 128 |
foreach ($posts as $post) { |
| 129 |
$grouped[$post->post_type][] = $post; |
| 130 |
} |
| 131 |
|
| 132 |
$type_labels = [ |
| 133 |
'page' => 'About', |
| 134 |
'post' => 'Blog', |
| 135 |
]; |
| 136 |
|
| 137 |
foreach ($grouped as $type => $items) { |
| 138 |
$label = isset($type_labels[$type]) ? $type_labels[$type] : ucfirst($type); |
| 139 |
$content .= '## ' . $label . "\n"; |
| 140 |
foreach ($items as $post) { |
| 141 |
$desc = $this->resolve_description($post); |
| 142 |
$line = '- [' . wp_strip_all_tags($post->post_title) . '](' . get_permalink($post) . ')'; |
| 143 |
if ($desc !== '') { |
| 144 |
$line .= ': ' . $desc; |
| 145 |
} |
| 146 |
$content .= $line . "\n"; |
| 147 |
} |
| 148 |
$content .= "\n"; |
| 149 |
} |
| 150 |
|
| 151 |
if (!empty($settings['llms_full_enabled'])) { |
| 152 |
$content .= "## Optional\n"; |
| 153 |
$content .= '- [llms-full.txt](' . site_url('/llms-full.txt') . '): Extended version with full content' . "\n"; |
| 154 |
} |
| 155 |
|
| 156 |
return $content; |
| 157 |
} |
| 158 |
|
| 159 |
/** |
| 160 |
* Build /llms-full.txt content. |
| 161 |
* |
| 162 |
* @return string |
| 163 |
*/ |
| 164 |
public function generate_full() |
| 165 |
{ |
| 166 |
if (!class_exists('Metasync_Html_To_Markdown')) { |
| 167 |
$converter_file = plugin_dir_path(dirname(__FILE__)) . 'includes/class-metasync-html-to-markdown.php'; |
| 168 |
if (file_exists($converter_file)) { |
| 169 |
require_once $converter_file; |
| 170 |
} |
| 171 |
} |
| 172 |
|
| 173 |
$settings = $this->get_settings(); |
| 174 |
$site_name = get_bloginfo('name'); |
| 175 |
$tagline = !empty($settings['custom_description']) |
| 176 |
? $settings['custom_description'] |
| 177 |
: get_bloginfo('description'); |
| 178 |
|
| 179 |
$content = '# ' . $site_name . "\n\n"; |
| 180 |
if (!empty($tagline)) { |
| 181 |
$content .= '> ' . $tagline . "\n\n"; |
| 182 |
} |
| 183 |
|
| 184 |
$posts = $this->query_posts_full($settings); |
| 185 |
$max_bytes = 1048576; // 1 MB |
| 186 |
|
| 187 |
foreach ($posts as $post) { |
| 188 |
$entry = "\n---\n\n"; |
| 189 |
if (class_exists('Metasync_Html_To_Markdown')) { |
| 190 |
$entry .= Metasync_Html_To_Markdown::convert_post($post->ID, [ |
| 191 |
'include_frontmatter' => true, |
| 192 |
'include_featured_image' => true, |
| 193 |
]); |
| 194 |
$entry .= "\n"; |
| 195 |
} else { |
| 196 |
$entry .= '# ' . $post->post_title . "\n\n"; |
| 197 |
$entry .= wp_strip_all_tags($post->post_content) . "\n"; |
| 198 |
} |
| 199 |
|
| 200 |
if (strlen($content) + strlen($entry) > $max_bytes) { |
| 201 |
break; |
| 202 |
} |
| 203 |
|
| 204 |
$content .= $entry; |
| 205 |
} |
| 206 |
|
| 207 |
return $content; |
| 208 |
} |
| 209 |
|
| 210 |
/** |
| 211 |
* Sync the robots.txt LLMs.txt reference when settings are saved. |
| 212 |
* |
| 213 |
* Called on both update_option_ and add_option_ hooks. For add_option the |
| 214 |
* hook passes ($option, $value) but we only need the second argument, which |
| 215 |
* is the new value in both cases. |
| 216 |
* |
| 217 |
* @param mixed $old_value Previous option value (or option name for add_option). |
| 218 |
* @param mixed $new_value New option value. |
| 219 |
*/ |
| 220 |
public function on_settings_changed( $old_value, $new_value ) { |
| 221 |
$robots_file = plugin_dir_path(dirname(__FILE__)) . 'robots-txt/class-metasync-robots-txt.php'; |
| 222 |
if (!class_exists('Metasync_Robots_Txt') && file_exists($robots_file)) { |
| 223 |
require_once $robots_file; |
| 224 |
} |
| 225 |
|
| 226 |
if (!class_exists('Metasync_Robots_Txt')) { |
| 227 |
return; |
| 228 |
} |
| 229 |
|
| 230 |
$robots = Metasync_Robots_Txt::get_instance(); |
| 231 |
$llms_url = site_url('/llms.txt'); |
| 232 |
|
| 233 |
if (!empty($new_value['enabled'])) { |
| 234 |
$robots->add_llms_txt_url($llms_url); |
| 235 |
} else { |
| 236 |
$robots->remove_llms_txt_url($llms_url); |
| 237 |
} |
| 238 |
} |
| 239 |
|
| 240 |
/** |
| 241 |
* Invalidate the cache only when the saved post's type is in the |
| 242 |
* configured LLMs.txt post_types list. |
| 243 |
* |
| 244 |
* @param int $post_id |
| 245 |
*/ |
| 246 |
public function maybe_invalidate_cache($post_id) |
| 247 |
{ |
| 248 |
$post = get_post($post_id); |
| 249 |
if (!$post) { |
| 250 |
return; |
| 251 |
} |
| 252 |
|
| 253 |
$saved = get_option(self::OPTION_KEY, []); |
| 254 |
$post_types = (is_array($saved) && !empty($saved['post_types']) && is_array($saved['post_types'])) |
| 255 |
? $saved['post_types'] |
| 256 |
: ['page', 'post']; |
| 257 |
|
| 258 |
if (in_array($post->post_type, $post_types, true)) { |
| 259 |
$this->invalidate_cache(); |
| 260 |
} |
| 261 |
} |
| 262 |
|
| 263 |
/** |
| 264 |
* Flush cached LLMs.txt content. |
| 265 |
*/ |
| 266 |
public function invalidate_cache() |
| 267 |
{ |
| 268 |
delete_transient(self::TRANSIENT_SHORT); |
| 269 |
delete_transient(self::TRANSIENT_FULL); |
| 270 |
} |
| 271 |
|
| 272 |
/** |
| 273 |
* Fetch posts that should appear in the listing. |
| 274 |
* |
| 275 |
* @param array $settings |
| 276 |
* @return WP_Post[] |
| 277 |
*/ |
| 278 |
private function query_posts($settings) |
| 279 |
{ |
| 280 |
$post_types = !empty($settings['post_types']) && is_array($settings['post_types']) |
| 281 |
? $settings['post_types'] |
| 282 |
: ['page', 'post']; |
| 283 |
|
| 284 |
$max_posts = isset($settings['max_posts']) ? (int) $settings['max_posts'] : 50; |
| 285 |
if ($max_posts < 1) { |
| 286 |
$max_posts = 50; |
| 287 |
} |
| 288 |
if ($max_posts > 500) { |
| 289 |
$max_posts = 500; |
| 290 |
} |
| 291 |
|
| 292 |
$excluded = !empty($settings['excluded_ids']) && is_array($settings['excluded_ids']) |
| 293 |
? array_map('absint', $settings['excluded_ids']) |
| 294 |
: []; |
| 295 |
|
| 296 |
$args = [ |
| 297 |
'post_type' => $post_types, |
| 298 |
'post_status' => 'publish', |
| 299 |
'posts_per_page' => $max_posts, |
| 300 |
'orderby' => 'modified', |
| 301 |
'order' => 'DESC', |
| 302 |
'no_found_rows' => true, |
| 303 |
'meta_query' => [ |
| 304 |
[ |
| 305 |
'relation' => 'OR', |
| 306 |
[ |
| 307 |
'key' => '_metasync_robots_index', |
| 308 |
'compare' => 'NOT EXISTS', |
| 309 |
], |
| 310 |
[ |
| 311 |
'key' => '_metasync_robots_index', |
| 312 |
'value' => 'noindex', |
| 313 |
'compare' => '!=', |
| 314 |
], |
| 315 |
], |
| 316 |
], |
| 317 |
]; |
| 318 |
|
| 319 |
if (!empty($excluded)) { |
| 320 |
$args['post__not_in'] = $excluded; |
| 321 |
} |
| 322 |
|
| 323 |
$query = new WP_Query($args); |
| 324 |
return $query->posts ?: []; |
| 325 |
} |
| 326 |
|
| 327 |
/** |
| 328 |
* Fetch posts for /llms-full.txt (uses the separate max_posts_full limit). |
| 329 |
* |
| 330 |
* @param array $settings |
| 331 |
* @return WP_Post[] |
| 332 |
*/ |
| 333 |
private function query_posts_full($settings) |
| 334 |
{ |
| 335 |
$override = $settings; |
| 336 |
$max = isset($settings['max_posts_full']) ? (int) $settings['max_posts_full'] : 25; |
| 337 |
if ($max < 1) { |
| 338 |
$max = 25; |
| 339 |
} |
| 340 |
if ($max > 500) { |
| 341 |
$max = 500; |
| 342 |
} |
| 343 |
$override['max_posts'] = $max; |
| 344 |
return $this->query_posts($override); |
| 345 |
} |
| 346 |
|
| 347 |
/** |
| 348 |
* Resolve the description for a post using the documented priority: |
| 349 |
* _metasync_seo_desc → _metasync_otto_description → manual excerpt → auto excerpt. |
| 350 |
* |
| 351 |
* @param WP_Post $post |
| 352 |
* @return string |
| 353 |
*/ |
| 354 |
private function resolve_description($post) |
| 355 |
{ |
| 356 |
$desc = get_post_meta($post->ID, '_metasync_seo_desc', true); |
| 357 |
if (!empty($desc)) { |
| 358 |
return $this->sanitize_description($desc); |
| 359 |
} |
| 360 |
|
| 361 |
$desc = get_post_meta($post->ID, '_metasync_otto_description', true); |
| 362 |
if (!empty($desc)) { |
| 363 |
return $this->sanitize_description($desc); |
| 364 |
} |
| 365 |
|
| 366 |
if (!empty($post->post_excerpt)) { |
| 367 |
return $this->sanitize_description($post->post_excerpt); |
| 368 |
} |
| 369 |
|
| 370 |
$auto = wp_trim_words(wp_strip_all_tags(strip_shortcodes($post->post_content)), 20, '…'); |
| 371 |
return $this->sanitize_description($auto); |
| 372 |
} |
| 373 |
|
| 374 |
/** |
| 375 |
* Collapse whitespace and normalise a description string for markdown output. |
| 376 |
* |
| 377 |
* @param string $desc |
| 378 |
* @return string |
| 379 |
*/ |
| 380 |
private function sanitize_description($desc) |
| 381 |
{ |
| 382 |
$desc = wp_strip_all_tags((string) $desc); |
| 383 |
$desc = preg_replace('/\s+/', ' ', $desc); |
| 384 |
return trim($desc); |
| 385 |
} |
| 386 |
|
| 387 |
/** |
| 388 |
* Get plugin settings with defaults merged. |
| 389 |
* |
| 390 |
* @return array |
| 391 |
*/ |
| 392 |
public function get_settings() |
| 393 |
{ |
| 394 |
$defaults = [ |
| 395 |
'enabled' => false, |
| 396 |
'post_types' => ['page', 'post'], |
| 397 |
'max_posts' => 50, |
| 398 |
'max_posts_full' => 25, |
| 399 |
'excluded_ids' => [], |
| 400 |
'custom_description' => '', |
| 401 |
'llms_full_enabled' => false, |
| 402 |
]; |
| 403 |
|
| 404 |
$saved = get_option(self::OPTION_KEY, []); |
| 405 |
if (!is_array($saved)) { |
| 406 |
$saved = []; |
| 407 |
} |
| 408 |
|
| 409 |
return array_merge($defaults, $saved); |
| 410 |
} |
| 411 |
|
| 412 |
/** |
| 413 |
* Detect whether another SEO plugin already handles /llms.txt. |
| 414 |
* |
| 415 |
* Caches the result in a short-lived transient so admin notices can |
| 416 |
* reflect the current state without re-checking on every request. |
| 417 |
* |
| 418 |
* @return bool |
| 419 |
*/ |
| 420 |
public function detect_plugin_conflict() |
| 421 |
{ |
| 422 |
$conflict = false; |
| 423 |
|
| 424 |
// Yoast SEO. |
| 425 |
if (class_exists('WPSEO_Options')) { |
| 426 |
$wpseo = get_option('wpseo'); |
| 427 |
if (is_array($wpseo) && !empty($wpseo['enable_llms_txt'])) { |
| 428 |
$conflict = true; |
| 429 |
} |
| 430 |
} |
| 431 |
|
| 432 |
// Rank Math. |
| 433 |
if (!$conflict && class_exists('RankMath')) { |
| 434 |
$rank_math_modules = get_option('rank_math_modules', []); |
| 435 |
if (is_array($rank_math_modules) && in_array('llms-txt', $rank_math_modules, true)) { |
| 436 |
$conflict = true; |
| 437 |
} |
| 438 |
if (!$conflict) { |
| 439 |
$rank_math_general = get_option('rank-math-options-general'); |
| 440 |
if (is_array($rank_math_general) && !empty($rank_math_general['llms_txt_enable'])) { |
| 441 |
$conflict = true; |
| 442 |
} |
| 443 |
} |
| 444 |
} |
| 445 |
|
| 446 |
// All in One SEO. |
| 447 |
if (!$conflict && function_exists('aioseo')) { |
| 448 |
try { |
| 449 |
$aioseo = aioseo(); |
| 450 |
if (isset($aioseo->options) && isset($aioseo->options->searchAppearance) |
| 451 |
&& isset($aioseo->options->searchAppearance->global) |
| 452 |
&& isset($aioseo->options->searchAppearance->global->llmsEnabled) |
| 453 |
&& $aioseo->options->searchAppearance->global->llmsEnabled) { |
| 454 |
$conflict = true; |
| 455 |
} |
| 456 |
} catch (\Throwable $e) { |
| 457 |
// Ignore – AIOSEO API changed or not fully loaded. |
| 458 |
} |
| 459 |
|
| 460 |
if (!$conflict) { |
| 461 |
$aioseo_raw = get_option('aioseo_options'); |
| 462 |
$aioseo_options = is_string($aioseo_raw) ? json_decode($aioseo_raw, true) : $aioseo_raw; |
| 463 |
if (is_array($aioseo_options) |
| 464 |
&& !empty($aioseo_options['searchAppearance']['global']['llmsEnabled'])) { |
| 465 |
$conflict = true; |
| 466 |
} |
| 467 |
} |
| 468 |
} |
| 469 |
|
| 470 |
if ($conflict) { |
| 471 |
set_transient(self::CONFLICT_TRANSIENT, true, HOUR_IN_SECONDS); |
| 472 |
} else { |
| 473 |
delete_transient(self::CONFLICT_TRANSIENT); |
| 474 |
} |
| 475 |
|
| 476 |
return $conflict; |
| 477 |
} |
| 478 |
} |
| 479 |
|
