| 1 |
<?php |
| 2 |
// If this file is called directly, abort. |
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
/** |
| 8 |
* SEO/front-end output functionality of the plugin. |
| 9 |
* |
| 10 |
* Handles all SEO meta tag output, structured data, robots directives, |
| 11 |
* and sitemap filtering for the public-facing side of the site. |
| 12 |
* |
| 13 |
* @link https://searchatlas.com |
| 14 |
* @since 1.0.0 |
| 15 |
* |
| 16 |
* @package Metasync |
| 17 |
* @subpackage Metasync/public |
| 18 |
* @author Engineering Team <[email protected]> |
| 19 |
*/ |
| 20 |
|
| 21 |
class Metasync_Seo_Output |
| 22 |
{ |
| 23 |
private $plugin_name; |
| 24 |
private $version; |
| 25 |
private $common; |
| 26 |
private $metasync_option_data; |
| 27 |
|
| 28 |
public function __construct($plugin_name, $version) |
| 29 |
{ |
| 30 |
$this->plugin_name = $plugin_name; |
| 31 |
$this->version = $version; |
| 32 |
$this->common = new Metasync_Common(); |
| 33 |
$this->metasync_option_data = Metasync::get_option('general'); |
| 34 |
|
| 35 |
// Filter permalink to use primary category (for canonical URLs and sitemaps) |
| 36 |
add_filter('post_link_category', array($this, 'filter_post_link_category_primary'), 10, 3); |
| 37 |
} |
| 38 |
|
| 39 |
/** |
| 40 |
* Filter the category used in post permalinks to respect the primary category. |
| 41 |
* Falls back to Yoast SEO or Rank Math primary category if MetaSync's is not set. |
| 42 |
* |
| 43 |
* @param WP_Term $cat The category to use in the permalink |
| 44 |
* @param array $cats Array of all categories associated with the post |
| 45 |
* @param WP_Post $post The post object |
| 46 |
* @return WP_Term The primary category term or the original category |
| 47 |
*/ |
| 48 |
public function filter_post_link_category_primary($cat, $cats, $post) |
| 49 |
{ |
| 50 |
// When the sitemap generator is collecting URLs on a %category% permalink |
| 51 |
// site, it primes a per-chunk prefetch map of the primary-category meta |
| 52 |
// keys (because update_post_meta_cache is disabled there to avoid memory |
| 53 |
// exhaustion on large catalogs). Read from that map first and only fall |
| 54 |
// back to get_post_meta() — which lazily fetches on a cache miss — when |
| 55 |
// the map has no entry for this post. Outside sitemap generation the map |
| 56 |
// is empty, so behaviour is unchanged. |
| 57 |
$prefetch = (class_exists('Metasync_Sitemap_Generator') && !empty(Metasync_Sitemap_Generator::$primary_category_prefetch)) |
| 58 |
? Metasync_Sitemap_Generator::$primary_category_prefetch |
| 59 |
: []; |
| 60 |
|
| 61 |
$primary_id = (int) self::primary_meta_lookup($post->ID, '_metasync_primary_category', $prefetch); |
| 62 |
|
| 63 |
// Fallback to Yoast SEO primary category |
| 64 |
if ($primary_id === 0 && defined('WPSEO_VERSION')) { |
| 65 |
$primary_id = (int) self::primary_meta_lookup($post->ID, '_yoast_wpseo_primary_category', $prefetch); |
| 66 |
} |
| 67 |
|
| 68 |
// Fallback to Rank Math primary category |
| 69 |
if ($primary_id === 0 && defined('RANK_MATH_VERSION')) { |
| 70 |
$primary_id = (int) self::primary_meta_lookup($post->ID, 'rank_math_primary_category', $prefetch); |
| 71 |
} |
| 72 |
|
| 73 |
if ($primary_id === 0 || empty($cats)) { |
| 74 |
return $cat; |
| 75 |
} |
| 76 |
|
| 77 |
foreach ($cats as $category) { |
| 78 |
if ((int) $category->term_id === $primary_id) { |
| 79 |
return $category; |
| 80 |
} |
| 81 |
} |
| 82 |
|
| 83 |
return $cat; |
| 84 |
} |
| 85 |
|
| 86 |
/** |
| 87 |
* Resolve a single primary-category meta value, preferring the sitemap |
| 88 |
* generator's per-chunk prefetch map (populated on %category% permalink |
| 89 |
* sites where the WP_Query post-meta cache prime is disabled) and falling |
| 90 |
* back to get_post_meta() when the map has no entry for the post/key. |
| 91 |
* |
| 92 |
* @param int $post_id Post ID. |
| 93 |
* @param string $key Meta key to resolve. |
| 94 |
* @param array $prefetch Map of post_id => [meta_key => value]. |
| 95 |
* @return string The scalar meta value (empty string when not set). |
| 96 |
*/ |
| 97 |
private static function primary_meta_lookup($post_id, $key, array $prefetch) |
| 98 |
{ |
| 99 |
$pid = (int) $post_id; |
| 100 |
if (isset($prefetch[$pid]) && is_array($prefetch[$pid]) && array_key_exists($key, $prefetch[$pid])) { |
| 101 |
return (string) $prefetch[$pid][$key]; |
| 102 |
} |
| 103 |
return (string) get_post_meta($post_id, $key, true); |
| 104 |
} |
| 105 |
|
| 106 |
function metasync_wp_robots_meta($robots) |
| 107 |
{ |
| 108 |
// Both robots features switched off — MetaSync emits no robots tag of its |
| 109 |
// own, so blanking core's directives here would leave the page with no |
| 110 |
// robots tag at all. Hand core's array back untouched. |
| 111 |
if (Metasync_Feature_Flags::robots_fully_disabled()) { |
| 112 |
return $robots; |
| 113 |
} |
| 114 |
// When the site is set to discourage search engines, core already |
| 115 |
// emits noindex; blanking every directive here would strip that and |
| 116 |
// make a discouraged site indexable. Leave the tags untouched. |
| 117 |
if ((int) get_option('blog_public', 1) === 0) { |
| 118 |
return $robots; |
| 119 |
} |
| 120 |
// On a singular view MetaSync only owns the tag when it actually resolves |
| 121 |
// a value. With one robots half switched off the resolved value can come |
| 122 |
// back empty — every directive the post had belonged to the disabled half |
| 123 |
// — and blanking core here would leave the page with no robots tag at all. |
| 124 |
if (function_exists('is_singular') && is_singular()) { |
| 125 |
$post_id = get_the_ID(); |
| 126 |
if ($post_id && $this->resolve_robots_value(get_post_meta($post_id)) === '') { |
| 127 |
return $robots; |
| 128 |
} |
| 129 |
} |
| 130 |
foreach ($robots as $key => $value) { |
| 131 |
$robots[$key] = false; |
| 132 |
} |
| 133 |
return $robots; |
| 134 |
} |
| 135 |
|
| 136 |
public function print_metatag($name, $value, $valueAttrib = "content", $nameAttrib = "name", $tagName = "meta") |
| 137 |
{ |
| 138 |
// Safety: extract string from array values (fixes legacy meta stored as |
| 139 |
// arrays). Recursive — a single reset() on a nested array |
| 140 |
// returns another array and printf would emit the literal "Array". |
| 141 |
if (is_array($value)) { |
| 142 |
$value = Metasync_Canonical_Sanitizer::extract_scalar($value); |
| 143 |
} |
| 144 |
|
| 145 |
if (empty($value)) |
| 146 |
return false; |
| 147 |
|
| 148 |
printf( |
| 149 |
"\t<%s %s=\"%s\" %s=\"%s\" />\n", |
| 150 |
esc_attr($tagName), |
| 151 |
esc_attr($nameAttrib), |
| 152 |
esc_attr($name), |
| 153 |
esc_attr($valueAttrib), |
| 154 |
esc_attr($value) |
| 155 |
); |
| 156 |
} |
| 157 |
|
| 158 |
/** |
| 159 |
* Check if the current page is an AMP page |
| 160 |
* |
| 161 |
* @return bool True if AMP page, false otherwise |
| 162 |
*/ |
| 163 |
public function is_amp_page() |
| 164 |
{ |
| 165 |
// Check if URL path contains /amp/ |
| 166 |
$current_url = $_SERVER['REQUEST_URI'] ?? ''; |
| 167 |
if (strpos($current_url, '/amp/') !== false) { |
| 168 |
return true; |
| 169 |
} |
| 170 |
|
| 171 |
// Check if URL ends with /amp |
| 172 |
if (preg_match('/\/amp\/?$/', $current_url)) { |
| 173 |
return true; |
| 174 |
} |
| 175 |
|
| 176 |
// Check if amp=1 query parameter is present |
| 177 |
if (isset($_GET['amp']) && $_GET['amp'] == '1') { |
| 178 |
return true; |
| 179 |
} |
| 180 |
|
| 181 |
// Check for other common AMP query parameters |
| 182 |
if (isset($_GET['amp']) && !empty($_GET['amp'])) { |
| 183 |
return true; |
| 184 |
} |
| 185 |
|
| 186 |
return false; |
| 187 |
} |
| 188 |
|
| 189 |
/** |
| 190 |
* Remove metasync_optimized attribute from head tag on AMP pages |
| 191 |
* This function uses output buffering to clean the head content |
| 192 |
*/ |
| 193 |
public function cleanup_amp_head_attribute() |
| 194 |
{ |
| 195 |
// Only run on AMP pages |
| 196 |
if (!$this->is_amp_page()) { |
| 197 |
return; |
| 198 |
} |
| 199 |
|
| 200 |
// Start output buffering to capture and modify the head content |
| 201 |
ob_start(function($buffer) { |
| 202 |
// Remove metasync_optimized attribute from head tag |
| 203 |
$cleaned_buffer = preg_replace('/(<head[^>]*)\s*metasync_optimized(?:="[^"]*")?([^>]*>)/i', '$1$2', $buffer); |
| 204 |
|
| 205 |
return $cleaned_buffer; |
| 206 |
}); |
| 207 |
} |
| 208 |
|
| 209 |
/** |
| 210 |
* End output buffering for AMP cleanup |
| 211 |
*/ |
| 212 |
public function end_amp_head_cleanup() |
| 213 |
{ |
| 214 |
// Only run on AMP pages |
| 215 |
if (!$this->is_amp_page()) { |
| 216 |
return; |
| 217 |
} |
| 218 |
|
| 219 |
// End output buffering |
| 220 |
if (ob_get_level()) { |
| 221 |
ob_end_flush(); |
| 222 |
} |
| 223 |
} |
| 224 |
|
| 225 |
public function hook_metasync_metatags() |
| 226 |
{ |
| 227 |
$conflict_handler = Metasync_SEO_Conflict_Handler::get_instance(); |
| 228 |
|
| 229 |
// On archive pages (category, tag, author, date, blog home) get_the_ID() |
| 230 |
// returns the first post in the loop — not the archive itself. Guard here so |
| 231 |
// we never use that first-post ID for canonical / og:url on archives. |
| 232 |
// Instead, emit the correct archive URL and return early. |
| 233 |
if (!is_singular()) { |
| 234 |
if (!$conflict_handler->has_active_seo_plugin()) { |
| 235 |
$archive_url = $this->get_archive_canonical_url(); |
| 236 |
if (!empty($archive_url)) { |
| 237 |
// The canonical link and og:url are owned by two different |
| 238 |
// feature switches, so each is emitted on its own terms. |
| 239 |
if (Metasync_Feature_Flags::is_enabled(Metasync_Feature_Flags::CANONICAL)) { |
| 240 |
printf("\t<link rel=\"canonical\" href=\"%s\" />\n", esc_url($archive_url)); |
| 241 |
} |
| 242 |
if (Metasync_Feature_Flags::is_enabled(Metasync_Feature_Flags::SOCIAL_OG)) { |
| 243 |
printf("\t<meta property=\"og:url\" content=\"%s\" />\n", esc_url($archive_url)); |
| 244 |
} |
| 245 |
} |
| 246 |
} |
| 247 |
return; |
| 248 |
} |
| 249 |
|
| 250 |
// Singular post/page: suppress WP core's canonical and let MetaSync handle it. |
| 251 |
// With the canonical feature switched off MetaSync emits none of its own, so |
| 252 |
// core's must be left in place rather than removed. |
| 253 |
if (!$conflict_handler->has_active_seo_plugin() |
| 254 |
&& Metasync_Feature_Flags::is_enabled(Metasync_Feature_Flags::CANONICAL)) { |
| 255 |
remove_action('wp_head', 'rel_canonical'); |
| 256 |
} |
| 257 |
|
| 258 |
$get_page_meta = get_post_meta(get_the_ID()); |
| 259 |
|
| 260 |
// When a third-party SEO plugin is active, only output our description |
| 261 |
// if MetaSync has an intentional value (OTTO or sidebar). Otherwise let |
| 262 |
// the other plugin handle it to avoid duplicates. |
| 263 |
$include_description = $conflict_handler->should_output_legacy_description(); |
| 264 |
|
| 265 |
$list_page_meta = array(); |
| 266 |
|
| 267 |
// When synced to an active SEO plugin, skip MetaSync's own |
| 268 |
// robots and description output — the native plugin handles it. |
| 269 |
$post_id = get_the_ID(); |
| 270 |
$is_synced_to_plugin = false; |
| 271 |
if ($post_id && $conflict_handler->has_active_seo_plugin()) { |
| 272 |
$sync_ts = get_post_meta($post_id, '_metasync_plugin_sync_ts', true); |
| 273 |
if (!empty($sync_ts)) { |
| 274 |
$is_synced_to_plugin = true; |
| 275 |
} |
| 276 |
} |
| 277 |
|
| 278 |
// Robots: resolve from both meta_robots (REST API) and metasync_common_robots (checkbox). |
| 279 |
// When a third-party SEO plugin is active: |
| 280 |
// - MetaSync has robots value → output ours, AIOSEO suppressed via filter |
| 281 |
// - MetaSync has NO value → skip, let the other plugin handle it |
| 282 |
// When no third-party plugin → always output if we have a value. |
| 283 |
if (!$is_synced_to_plugin) { |
| 284 |
$robots_value = $this->resolve_robots_value($get_page_meta); |
| 285 |
if (!empty($robots_value)) { |
| 286 |
if (!$conflict_handler->has_active_seo_plugin()) { |
| 287 |
$list_page_meta['robots'] = $robots_value; |
| 288 |
} else { |
| 289 |
// Third-party active but MetaSync has intentional robots — output ours |
| 290 |
// (the conflict handler filter suppresses the other plugin's robots tag) |
| 291 |
if ($post_id && $conflict_handler->metasync_has_robots($post_id)) { |
| 292 |
$list_page_meta['robots'] = $robots_value; |
| 293 |
} |
| 294 |
} |
| 295 |
} |
| 296 |
} |
| 297 |
|
| 298 |
if ($include_description) { |
| 299 |
$list_page_meta['description'] = $get_page_meta['meta_description'][0] ?? ''; |
| 300 |
} |
| 301 |
|
| 302 |
// Note: enable_metadesc is always enabled by default - no check needed |
| 303 |
|
| 304 |
$getSearchEngineOptions = Metasync::get_option('searchengines'); |
| 305 |
$keysSearchEngines = [ |
| 306 |
'bing_site_verification' => 'msvalidate.01', |
| 307 |
'baidu_site_verification' => 'baidu-site-verification', |
| 308 |
'alexa_site_verification' => 'alexaVerifyID', |
| 309 |
'yandex_site_verification' => 'yandex-verification', |
| 310 |
'google_site_verification' => 'google-site-verification', |
| 311 |
'pinterest_site_verification' => 'p:domain_verify', |
| 312 |
'norton_save_site_verification' => 'norton-safeweb-site-verification', |
| 313 |
]; |
| 314 |
|
| 315 |
$post = get_post(get_the_ID()); |
| 316 |
if (empty($post)) |
| 317 |
return; |
| 318 |
|
| 319 |
# Keywords: render persisted OTTO focus keyword as <meta name="keywords"> |
| 320 |
$focus_keyword = get_post_meta($post->ID, '_metasync_focus_keyword', true); |
| 321 |
if (!empty($focus_keyword)) { |
| 322 |
$list_page_meta['keywords'] = sanitize_text_field($focus_keyword); |
| 323 |
} |
| 324 |
|
| 325 |
# Canonical: output <link rel="canonical"> only when no third-party SEO plugin |
| 326 |
# handles it, to avoid duplicate canonical tags in the page source. |
| 327 |
if (!$conflict_handler->has_active_seo_plugin() |
| 328 |
&& Metasync_Feature_Flags::is_enabled(Metasync_Feature_Flags::CANONICAL)) { |
| 329 |
$list_page_meta['canonical'] = $this->get_canonical_url($post); |
| 330 |
} |
| 331 |
|
| 332 |
# $post_text = wp_trim_words(get_the_content(), 30, ''); |
| 333 |
|
| 334 |
# Check if the post has content, then apply WordPress content filters |
| 335 |
# $post_content = !empty($post->post_content) ? apply_filters('the_content', $post->post_content) : ''; |
| 336 |
# $post_text = wp_trim_words($post_content, 30, ''); |
| 337 |
|
| 338 |
/** |
| 339 |
* Extract post content safely for meta descriptions |
| 340 |
* This solution addresses three critical issues: |
| 341 |
* 1. Timber compatibility - works without WordPress loop |
| 342 |
* 2. Plugin conflicts - prevents shortcode execution (e.g., Hostify-booking function redeclaration) |
| 343 |
* 3. Performance - lightweight content extraction for meta tags |
| 344 |
*/ |
| 345 |
|
| 346 |
$post_text = ''; |
| 347 |
|
| 348 |
# Get content safely without executing shortcodes |
| 349 |
$content = get_the_content(null, false, $post); |
| 350 |
|
| 351 |
# Validate content exists |
| 352 |
if (!empty($content)) { |
| 353 |
# Clean content for meta description - multi-layer approach for reliability |
| 354 |
$content = strip_shortcodes($content); # Remove shortcode tags (e.g., [hostify-booking]) |
| 355 |
$content = wp_strip_all_tags($content); # Remove HTML tags |
| 356 |
$content = preg_replace('/\s+/', ' ', $content); # Normalize whitespace |
| 357 |
$post_text = wp_trim_words(trim($content), 30, ''); |
| 358 |
} |
| 359 |
|
| 360 |
$site_info = Metasync::get_option('optimal_settings')['site_info'] ?? []; |
| 361 |
|
| 362 |
$facebook_page_url = Metasync::get_option('social_meta')['facebook_page_url'] ?? ''; |
| 363 |
$facebook_authorship = Metasync::get_option('social_meta')['facebook_authorship'] ?? ''; |
| 364 |
$facebook_admin = Metasync::get_option('social_meta')['facebook_admin'] ?? ''; |
| 365 |
|
| 366 |
$twitter_username = Metasync::get_option('social_meta')['twitter_username'] ?? ''; |
| 367 |
|
| 368 |
$image = []; |
| 369 |
$image_mime_type = ''; |
| 370 |
|
| 371 |
// Resolve the Open Graph image from attachment metadata only — no disk |
| 372 |
// reads, no HTTP calls. This path is READ-ONLY: it must never mutate post |
| 373 |
// meta while rendering wp_head. Media-offload plugins (WP Stateless, WP |
| 374 |
// Offload Media / S3) store the actual files in remote object storage, so |
| 375 |
// the local file is legitimately absent. A missing local file is NOT proof |
| 376 |
// of an orphaned attachment and must never trigger deletion of the featured |
| 377 |
// image association. |
| 378 |
if ($post) { |
| 379 |
$image_id = get_post_thumbnail_id($post->ID); |
| 380 |
if ($image_id) { |
| 381 |
$og_image = $this->build_og_image_from_attachment($image_id); |
| 382 |
if ($og_image) { |
| 383 |
$image = [$og_image['url'], $og_image['width'], $og_image['height']]; |
| 384 |
$image_mime_type = $og_image['mime']; |
| 385 |
} |
| 386 |
} |
| 387 |
} |
| 388 |
|
| 389 |
// Fallback to site default image if post has no featured image. |
| 390 |
if (empty($image) && $site_info && !empty($site_info['social_share_image'])) { |
| 391 |
$og_image = $this->build_og_image_from_attachment($site_info['social_share_image']); |
| 392 |
if ($og_image) { |
| 393 |
$image = [$og_image['url'], $og_image['width'], $og_image['height']]; |
| 394 |
$image_mime_type = $og_image['mime']; |
| 395 |
} |
| 396 |
} |
| 397 |
|
| 398 |
|
| 399 |
// When a third-party SEO plugin handles OG/Twitter, omit our description |
| 400 |
// from those tags to avoid duplicates. |
| 401 |
$og_description = ($include_description) ? ($post_text ?? '') : ''; |
| 402 |
|
| 403 |
$ogMetaKeys = [ |
| 404 |
'og:locale' => get_locale(), |
| 405 |
'og:type' => 'article', |
| 406 |
'og:title' => $post->post_title . ' - ' . get_bloginfo('name'), |
| 407 |
'og:description' => $og_description, |
| 408 |
'og:url' => $this->get_canonical_url($post), |
| 409 |
'og:site_name' => get_bloginfo('name'), |
| 410 |
'og:updated_time' => $post->post_modified, |
| 411 |
'og:image' => $image ? $image[0] : '', |
| 412 |
'og:image:width' => $image ? $image[1] : '', |
| 413 |
'og:image:height' => $image ? $image[2] : '', |
| 414 |
'og:image:type' => $image ? $image_mime_type : '', |
| 415 |
'og:image:alt' => $image ? $post->post_title : '', |
| 416 |
]; |
| 417 |
|
| 418 |
$facebookMetaKeys = [ |
| 419 |
'article:publisher' => $facebook_page_url && !filter_var($facebook_page_url, FILTER_VALIDATE_URL) ? 'https://' . $facebook_page_url : $facebook_page_url, |
| 420 |
'article:author' => $facebook_authorship && !filter_var($facebook_authorship, FILTER_VALIDATE_URL) ? 'https://' . $facebook_authorship : $facebook_authorship, |
| 421 |
'fb:admins' => $facebook_admin, |
| 422 |
]; |
| 423 |
|
| 424 |
$twitter_card_type = Metasync::get_option('twitter_card_type') ?? []; |
| 425 |
|
| 426 |
$twitterMetaKeys = [ |
| 427 |
'twitter:card' => $twitter_card_type ? $twitter_card_type : 'summary_large_image', |
| 428 |
'twitter:title' => $post->post_title . ' - ' . get_bloginfo('name'), |
| 429 |
'twitter:site' => $twitter_username ? '@' . $twitter_username : '', |
| 430 |
'twitter:creator' => $twitter_username ? '@' . $twitter_username : '', |
| 431 |
'twitter:description' => $og_description, |
| 432 |
'twitter:image' => $image ? $image[0] : '', |
| 433 |
]; |
| 434 |
|
| 435 |
// echo "\t<!-- MetaSync metadata -->\n"; |
| 436 |
|
| 437 |
foreach ($list_page_meta as $item => $value) { |
| 438 |
if ($item == 'canonical') { |
| 439 |
$this->print_metatag($item, $value, 'href', 'rel', 'link'); |
| 440 |
continue; |
| 441 |
} |
| 442 |
$this->print_metatag($item, $value); |
| 443 |
} |
| 444 |
|
| 445 |
if ($getSearchEngineOptions !== null) { // check if searchengine verification options are set |
| 446 |
foreach ($keysSearchEngines as $optionKey => $metaKey) { |
| 447 |
$this->print_metatag($metaKey, $getSearchEngineOptions[$optionKey]); |
| 448 |
} |
| 449 |
} |
| 450 |
|
| 451 |
if ($post) { |
| 452 |
|
| 453 |
// When OTTO is active AND has OG/Twitter data for this post, skip |
| 454 |
// legacy output. For dynamic pixel injection (where specific OTTO OG |
| 455 |
// meta is empty), the buffer-level dedup in |
| 456 |
// Otto_html_class::deduplicate_og_twitter_tags() handles cleanup. |
| 457 |
$otto_has_og = false; |
| 458 |
$otto_has_twitter = false; |
| 459 |
if (class_exists('Metasync_Otto_Config') && Metasync_Otto_Config::is_otto_enabled()) { |
| 460 |
$otto_has_og = !empty(get_post_meta($post->ID, '_metasync_otto_og_title', true)) |
| 461 |
|| !empty(get_post_meta($post->ID, '_metasync_otto_og_description', true)); |
| 462 |
$otto_has_twitter = !empty(get_post_meta($post->ID, '_metasync_otto_twitter_title', true)) |
| 463 |
|| !empty(get_post_meta($post->ID, '_metasync_otto_twitter_description', true)); |
| 464 |
} |
| 465 |
|
| 466 |
// When the canonical emitter (Metasync_OpenGraph::output_opengraph_tags, wp_head |
| 467 |
// priority 5) will emit for this post, suppress these legacy og:*/twitter:* blocks |
| 468 |
// so a MetaSync-only page outputs exactly one OG set and one Twitter set, preferring |
| 469 |
// the persisted/OTTO values that emitter resolves. The Facebook block |
| 470 |
// (article:publisher / article:author / fb:admins) is NOT gated on it — the canonical |
| 471 |
// emitter never outputs those tags, so they never duplicate. |
| 472 |
$og2_will_emit = apply_filters('metasync_opengraph_will_emit', false); |
| 473 |
|
| 474 |
// When a third-party SEO plugin (Yoast / Rank Math / AIOSEO) is active it owns the |
| 475 |
// social tags — suppress ALL legacy Facebook/OG/Twitter output so MetaSync never |
| 476 |
// duplicates the other plugin's og:*/twitter:*/article:* tags. |
| 477 |
$has_seo_plugin = $conflict_handler->has_active_seo_plugin(); |
| 478 |
|
| 479 |
// The Social Media & Open Graph feature owns every tag emitted below — |
| 480 |
// Facebook, Open Graph and Twitter alike. Switched off, this legacy |
| 481 |
// fallback must stay silent just like the canonical emitter does. |
| 482 |
$social_enabled = Metasync_Feature_Flags::is_enabled(Metasync_Feature_Flags::SOCIAL_OG); |
| 483 |
|
| 484 |
$common_meta_settings = Metasync::get_option('common_meta_settings') ?? []; |
| 485 |
|
| 486 |
if ($social_enabled && !$otto_has_og && !$has_seo_plugin && isset($common_meta_settings['facebook_meta_tags'])) { |
| 487 |
foreach ($facebookMetaKeys as $metaKey => $metaValue) { |
| 488 |
$this->print_metatag($metaKey, $metaValue, 'content', 'property'); |
| 489 |
} |
| 490 |
} |
| 491 |
|
| 492 |
if ($social_enabled && !$otto_has_og && !$og2_will_emit && !$has_seo_plugin && isset($common_meta_settings['open_graph_meta_tags'])) { |
| 493 |
foreach ($ogMetaKeys as $metaKey => $metaValue) { |
| 494 |
$this->print_metatag($metaKey, $metaValue, 'content', 'property'); |
| 495 |
} |
| 496 |
} |
| 497 |
if ($social_enabled && !$otto_has_twitter && !$og2_will_emit && !$has_seo_plugin && isset($common_meta_settings['twitter_meta_tags'])) { |
| 498 |
foreach ($twitterMetaKeys as $metaKey => $metaValue) { |
| 499 |
$this->print_metatag($metaKey, $metaValue, 'content', 'name'); |
| 500 |
} |
| 501 |
} |
| 502 |
} |
| 503 |
} |
| 504 |
|
| 505 |
/** |
| 506 |
* Build Open Graph image data for an attachment using database metadata only. |
| 507 |
* |
| 508 |
* Intentionally avoids `file_exists()`/disk access and any HTTP validation so |
| 509 |
* it works for media-offload setups (WP Stateless, WP Offload Media / S3) where |
| 510 |
* the file lives in remote object storage rather than on the local filesystem. |
| 511 |
* It is strictly read-only and never mutates post or attachment meta. |
| 512 |
* |
| 513 |
* The image URL is resolved via `wp_get_attachment_url()`, which offload plugins |
| 514 |
* filter to the correct remote URL; it falls back to building the URL from the |
| 515 |
* upload directory and stored metadata when needed. |
| 516 |
* |
| 517 |
* @param int|string $attachment_id Attachment ID. |
| 518 |
* @return array|null ['url' => string, 'width' => int, 'height' => int, 'mime' => string] or null. |
| 519 |
*/ |
| 520 |
private function build_og_image_from_attachment($attachment_id) |
| 521 |
{ |
| 522 |
$attachment_id = (int) $attachment_id; |
| 523 |
if ($attachment_id <= 0) { |
| 524 |
return null; |
| 525 |
} |
| 526 |
|
| 527 |
$attachment = get_post($attachment_id); |
| 528 |
if (!$attachment || $attachment->post_type !== 'attachment') { |
| 529 |
return null; |
| 530 |
} |
| 531 |
|
| 532 |
// Read dimensions from stored metadata — no disk or network access. |
| 533 |
$metadata = wp_get_attachment_metadata($attachment_id); |
| 534 |
$width = (is_array($metadata) && !empty($metadata['width'])) ? (int) $metadata['width'] : 0; |
| 535 |
$height = (is_array($metadata) && !empty($metadata['height'])) ? (int) $metadata['height'] : 0; |
| 536 |
|
| 537 |
// Prefer wp_get_attachment_url() so offload plugins (WP Stateless / S3) |
| 538 |
// return the correct remote URL; fall back to upload dir + metadata file. |
| 539 |
$image_url = wp_get_attachment_url($attachment_id); |
| 540 |
if (!$image_url && is_array($metadata) && !empty($metadata['file'])) { |
| 541 |
$upload_dir = wp_upload_dir(); |
| 542 |
if (empty($upload_dir['error'])) { |
| 543 |
$image_url = trailingslashit($upload_dir['baseurl']) . ltrim($metadata['file'], '/'); |
| 544 |
} |
| 545 |
} |
| 546 |
|
| 547 |
if (!$image_url) { |
| 548 |
return null; |
| 549 |
} |
| 550 |
|
| 551 |
// Determine MIME type from the file extension (safe, no HTTP calls). |
| 552 |
$mime_types = [ |
| 553 |
'jpg' => 'image/jpeg', |
| 554 |
'jpeg' => 'image/jpeg', |
| 555 |
'png' => 'image/png', |
| 556 |
'gif' => 'image/gif', |
| 557 |
'webp' => 'image/webp', |
| 558 |
'svg' => 'image/svg+xml', |
| 559 |
]; |
| 560 |
$file_ext = strtolower(pathinfo(parse_url($image_url, PHP_URL_PATH) ?: $image_url, PATHINFO_EXTENSION)); |
| 561 |
$image_mime_type = $mime_types[$file_ext] ?? 'image/jpeg'; |
| 562 |
|
| 563 |
return [ |
| 564 |
'url' => $image_url, |
| 565 |
'width' => $width, |
| 566 |
'height' => $height, |
| 567 |
'mime' => $image_mime_type, |
| 568 |
]; |
| 569 |
} |
| 570 |
|
| 571 |
public function add_ld_json() |
| 572 |
{ |
| 573 |
$post = get_post(get_the_ID()); |
| 574 |
if (empty($post)) |
| 575 |
return; |
| 576 |
|
| 577 |
$site_info = Metasync::get_option('optimal_settings')['site_info'] ?? ''; |
| 578 |
|
| 579 |
$site_logo_id = $site_info['google_logo'] ?? ''; |
| 580 |
$custom_logo_id = get_theme_mod('custom_logo'); |
| 581 |
$logo_id = $custom_logo_id != '' ? $custom_logo_id : $site_logo_id; |
| 582 |
|
| 583 |
$site_image_id = $site_info['social_share_image'] ?? ''; |
| 584 |
$post_thumbnail_id = get_post_thumbnail_id($post->ID); |
| 585 |
$thumbnail_id = $post_thumbnail_id > 0 ? $post_thumbnail_id : $site_image_id; |
| 586 |
|
| 587 |
$schema = array( |
| 588 |
'@context' => "http://schema.org", |
| 589 |
'@type' => "Article", |
| 590 |
'headline' => $post->post_title ?? '', |
| 591 |
'image' => wp_get_attachment_image_url($thumbnail_id, 'full'), |
| 592 |
'url' => get_permalink(), |
| 593 |
'datePublished' => $post->post_modified, |
| 594 |
'author' => array( |
| 595 |
'@type' => "Person", |
| 596 |
'name' => get_the_author_meta('display_name', $post->post_author), |
| 597 |
'url' => get_author_posts_url($post->post_author), |
| 598 |
), |
| 599 |
'publisher' => array( |
| 600 |
'@type' => "Organization", |
| 601 |
'name' => get_bloginfo('name') ?? '', |
| 602 |
'url' => get_site_url(), |
| 603 |
'logo' => array( |
| 604 |
'@type' => "ImageObject", |
| 605 |
'url' => wp_get_attachment_image_url($logo_id, 'full'), |
| 606 |
) |
| 607 |
) |
| 608 |
); |
| 609 |
|
| 610 |
return $schema; |
| 611 |
} |
| 612 |
|
| 613 |
/** |
| 614 |
* Output the LocalBusiness / Organization / Person JSON-LD built from the |
| 615 |
* Local Business settings page (metasync_options['localseo']). |
| 616 |
* |
| 617 |
* Hooked to wp_head so the saved Local Business data appears in the |
| 618 |
* front-end page source as valid schema.org structured data. Emits nothing |
| 619 |
* on admin screens, feeds, or when the Local Business page is unconfigured. |
| 620 |
*/ |
| 621 |
public function output_local_business_schema() |
| 622 |
{ |
| 623 |
if (is_admin() || is_feed()) { |
| 624 |
return; |
| 625 |
} |
| 626 |
|
| 627 |
// The Schema Markup switch covers every JSON-LD block MetaSync owns, |
| 628 |
// site-wide Local SEO markup included — not just the per-post editor box. |
| 629 |
if (Metasync_Feature_Flags::is_disabled(Metasync_Feature_Flags::SCHEMA)) { |
| 630 |
return; |
| 631 |
} |
| 632 |
|
| 633 |
$schema = $this->build_local_business_schema(); |
| 634 |
if (empty($schema)) { |
| 635 |
return; |
| 636 |
} |
| 637 |
|
| 638 |
$json = metasync_safe_json_ld_encode($schema, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE); |
| 639 |
if (empty($json)) { |
| 640 |
return; |
| 641 |
} |
| 642 |
|
| 643 |
echo "\t<script type=\"application/ld+json\">" . $json . "</script>\n"; |
| 644 |
} |
| 645 |
|
| 646 |
/** |
| 647 |
* Build the LocalBusiness structured-data entity from the saved localseo |
| 648 |
* option. Returns null when the page has never been configured (no Person/ |
| 649 |
* Organization type selected), so unconfigured sites emit no empty schema. |
| 650 |
* |
| 651 |
* @param array|null $options Saved localseo options; when null they are |
| 652 |
* read from metasync_options['localseo']. |
| 653 |
* @return array|null JSON-LD entity, or null when nothing is configured. |
| 654 |
*/ |
| 655 |
public function build_local_business_schema($options = null) |
| 656 |
{ |
| 657 |
if ($options === null) { |
| 658 |
$options = Metasync::get_option('localseo'); |
| 659 |
} |
| 660 |
if (empty($options) || !is_array($options)) { |
| 661 |
return null; |
| 662 |
} |
| 663 |
|
| 664 |
$type = $options['local_seo_person_organization'] ?? ''; |
| 665 |
if ($type !== 'Organization' && $type !== 'Person') { |
| 666 |
return null; |
| 667 |
} |
| 668 |
|
| 669 |
$url = !empty($options['local_seo_url']) ? $options['local_seo_url'] : get_home_url(); |
| 670 |
|
| 671 |
$entity = array( |
| 672 |
'@context' => 'https://schema.org', |
| 673 |
'@type' => '', |
| 674 |
'name' => '', |
| 675 |
'url' => $url, |
| 676 |
); |
| 677 |
|
| 678 |
if (!empty($options['local_seo_email'])) { |
| 679 |
$entity['email'] = $options['local_seo_email']; |
| 680 |
} |
| 681 |
|
| 682 |
if ($type === 'Person') { |
| 683 |
$entity = $this->local_business_person($entity, $options); |
| 684 |
} else { |
| 685 |
$entity = $this->local_business_organization($entity, $options); |
| 686 |
} |
| 687 |
|
| 688 |
return $entity; |
| 689 |
} |
| 690 |
|
| 691 |
/** |
| 692 |
* Populate an Organization / LocalBusiness entity. |
| 693 |
* |
| 694 |
* @param array $entity Base JSON-LD entity. |
| 695 |
* @param array $options Saved localseo options. |
| 696 |
* @return array |
| 697 |
*/ |
| 698 |
private function local_business_organization($entity, $options) |
| 699 |
{ |
| 700 |
$name = (string) ($options['local_seo_name'] ?? ''); |
| 701 |
$entity['name'] = $name !== '' ? $name : (string) get_bloginfo('name'); |
| 702 |
|
| 703 |
// The Business Type field stores a human-readable label (e.g. "Local |
| 704 |
// Business", "Art Gallery"). schema.org @type tokens have no spaces, so |
| 705 |
// collapse them ("Local Business" -> "LocalBusiness") to stay valid. |
| 706 |
// A truthy value can never be the string '0', which is itself falsy, so |
| 707 |
// the truthiness check alone already covers the "0" case. |
| 708 |
$business_type = $options['local_seo_business_type'] ?? ''; |
| 709 |
$schema_type = $business_type |
| 710 |
? str_replace(' ', '', $business_type) |
| 711 |
: 'Organization'; |
| 712 |
$entity['@type'] = $schema_type; |
| 713 |
|
| 714 |
if (!empty($options['local_seo_logo'])) { |
| 715 |
$logo_url = $this->resolve_local_seo_logo_url($options['local_seo_logo'], 'full'); |
| 716 |
if ($logo_url) { |
| 717 |
$entity['logo'] = $logo_url; |
| 718 |
} |
| 719 |
} |
| 720 |
|
| 721 |
if (!empty($options['local_seo_price_range'])) { |
| 722 |
$entity['priceRange'] = $options['local_seo_price_range']; |
| 723 |
} |
| 724 |
|
| 725 |
$this->local_business_add_geo($entity, $options); |
| 726 |
$this->local_business_add_address($entity, $options); |
| 727 |
$this->local_business_add_contact_points($entity, $options); |
| 728 |
$this->local_business_add_hours($entity, $options); |
| 729 |
|
| 730 |
return $entity; |
| 731 |
} |
| 732 |
|
| 733 |
/** |
| 734 |
* Populate a Person entity. |
| 735 |
* |
| 736 |
* @param array $entity Base JSON-LD entity. |
| 737 |
* @param array $options Saved localseo options. |
| 738 |
* @return array |
| 739 |
*/ |
| 740 |
private function local_business_person($entity, $options) |
| 741 |
{ |
| 742 |
$name = (string) ($options['local_seo_name'] ?? ''); |
| 743 |
$entity['@type'] = 'Person'; |
| 744 |
$entity['name'] = $name !== '' ? $name : (string) get_bloginfo('name'); |
| 745 |
|
| 746 |
if (!empty($options['local_seo_phone'])) { |
| 747 |
$entity['telephone'] = $options['local_seo_phone']; |
| 748 |
} |
| 749 |
|
| 750 |
if (!empty($options['local_seo_logo'])) { |
| 751 |
$logo_url = $this->resolve_local_seo_logo_url($options['local_seo_logo'], 'full'); |
| 752 |
if ($logo_url) { |
| 753 |
$entity['image'] = $logo_url; |
| 754 |
} |
| 755 |
} |
| 756 |
|
| 757 |
$this->local_business_add_geo($entity, $options); |
| 758 |
$this->local_business_add_address($entity, $options); |
| 759 |
|
| 760 |
return $entity; |
| 761 |
} |
| 762 |
|
| 763 |
/** |
| 764 |
* Resolve the stored Local Business logo to a publishable image URL. |
| 765 |
* |
| 766 |
* The setting holds either a media-library attachment ID or a raw URL. |
| 767 |
* Rows saved by the legacy sanitizer hold the corrupted form of an |
| 768 |
* ID ("http://45589"), which is neither numeric nor a resolvable URL, and |
| 769 |
* would otherwise be published verbatim as broken structured data. Repair |
| 770 |
* that shape first, then resolve the ID or trust the URL as stored. |
| 771 |
* |
| 772 |
* @param mixed $logo Stored logo value. |
| 773 |
* @param string $size Registered image size to request for an attachment. |
| 774 |
* @return string|false Resolved image URL, or false when there is none. |
| 775 |
*/ |
| 776 |
private function resolve_local_seo_logo_url($logo, $size = 'full') |
| 777 |
{ |
| 778 |
$logo = metasync_repair_scheme_prefixed_media_id($logo); |
| 779 |
|
| 780 |
if (is_numeric($logo)) { |
| 781 |
$url = wp_get_attachment_image_url($logo, $size); |
| 782 |
return $url ? $url : false; |
| 783 |
} |
| 784 |
|
| 785 |
return $logo ? $logo : false; |
| 786 |
} |
| 787 |
|
| 788 |
/** |
| 789 |
* Add a PostalAddress from the saved address sub-array. |
| 790 |
*/ |
| 791 |
private function local_business_add_address(&$entity, $options) |
| 792 |
{ |
| 793 |
$address = $options['address'] ?? array(); |
| 794 |
if (empty($address) || !is_array($address)) { |
| 795 |
return; |
| 796 |
} |
| 797 |
|
| 798 |
$street = $address['street'] ?? ''; |
| 799 |
$locality = $address['locality'] ?? ''; |
| 800 |
$region = $address['region'] ?? ''; |
| 801 |
$postal = $address['postalcode'] ?? ''; |
| 802 |
$country = $address['country'] ?? ''; |
| 803 |
|
| 804 |
// Skip entirely if every address field is empty. |
| 805 |
if ($street === '' && $locality === '' && $region === '' && $postal === '' && $country === '') { |
| 806 |
return; |
| 807 |
} |
| 808 |
|
| 809 |
$entity['address'] = array( |
| 810 |
'@type' => 'PostalAddress', |
| 811 |
'streetAddress' => $street, |
| 812 |
'addressLocality' => $locality, |
| 813 |
'addressRegion' => $region, |
| 814 |
'postalCode' => $postal, |
| 815 |
'addressCountry' => $country, |
| 816 |
); |
| 817 |
} |
| 818 |
|
| 819 |
/** |
| 820 |
* Add GeoCoordinates + hasMap from "lat,lng" saved coordinates. |
| 821 |
*/ |
| 822 |
private function local_business_add_geo(&$entity, $options) |
| 823 |
{ |
| 824 |
$coords = $options['local_seo_geo_coordinates'] ?? ''; |
| 825 |
if (empty($coords)) { |
| 826 |
return; |
| 827 |
} |
| 828 |
|
| 829 |
// explode() always yields at least one element, so only the second index |
| 830 |
// needs an existence check (a value with no comma has no longitude). |
| 831 |
$geo = array_map('trim', explode(',', $coords)); |
| 832 |
if (!isset($geo[1]) || $geo[0] === '' || $geo[1] === '') { |
| 833 |
return; |
| 834 |
} |
| 835 |
|
| 836 |
$entity['geo'] = array( |
| 837 |
'@type' => 'GeoCoordinates', |
| 838 |
'latitude' => $geo[0], |
| 839 |
'longitude' => $geo[1], |
| 840 |
); |
| 841 |
$entity['hasMap'] = 'https://www.google.com/maps/search/?api=1&query=' . rawurlencode($geo[0] . ',' . $geo[1]); |
| 842 |
} |
| 843 |
|
| 844 |
/** |
| 845 |
* Add ContactPoint entries from the parallel phonetype[] / phonenumber[] arrays. |
| 846 |
*/ |
| 847 |
private function local_business_add_contact_points(&$entity, $options) |
| 848 |
{ |
| 849 |
$types = $options['phonetype'] ?? ''; |
| 850 |
$numbers = $options['phonenumber'] ?? ''; |
| 851 |
if (!is_array($types) || !is_array($numbers) || empty($types) || empty($numbers)) { |
| 852 |
return; |
| 853 |
} |
| 854 |
|
| 855 |
$contacts = array(); |
| 856 |
foreach ($types as $i => $contact_type) { |
| 857 |
$number = $numbers[$i] ?? ''; |
| 858 |
if ($contact_type === '' || $number === '') { |
| 859 |
continue; |
| 860 |
} |
| 861 |
$contacts[] = array( |
| 862 |
'@type' => 'ContactPoint', |
| 863 |
'telephone' => $number, |
| 864 |
'contactType' => $contact_type, |
| 865 |
); |
| 866 |
} |
| 867 |
|
| 868 |
if (!empty($contacts)) { |
| 869 |
$entity['contactPoint'] = $contacts; |
| 870 |
} |
| 871 |
} |
| 872 |
|
| 873 |
/** |
| 874 |
* Add openingHours from the parallel days[] / times[] arrays. |
| 875 |
*/ |
| 876 |
private function local_business_add_hours(&$entity, $options) |
| 877 |
{ |
| 878 |
$days = $options['days'] ?? ''; |
| 879 |
$times = $options['times'] ?? ''; |
| 880 |
if (!is_array($days) || !is_array($times) || empty($days) || empty($times)) { |
| 881 |
return; |
| 882 |
} |
| 883 |
|
| 884 |
$hours = array(); |
| 885 |
foreach ($days as $i => $day) { |
| 886 |
$time = $times[$i] ?? ''; |
| 887 |
if ($day === '' || $time === '') { |
| 888 |
continue; |
| 889 |
} |
| 890 |
$hours[] = $day . ' ' . $time; |
| 891 |
} |
| 892 |
|
| 893 |
if (!empty($hours)) { |
| 894 |
$entity['openingHours'] = $hours; |
| 895 |
} |
| 896 |
} |
| 897 |
|
| 898 |
/* |
| 899 |
* This will hide the title on single posts and pages |
| 900 |
* if they were created with the "metasync" system. |
| 901 |
* Passes default values for $title and $id to avoid errors. |
| 902 |
*/ |
| 903 |
public function hide_title_on_otto_pages($title = '', $id = null){ |
| 904 |
|
| 905 |
# Return title immediately if $id or $title is not provided |
| 906 |
if (empty($id) || empty($title)) { |
| 907 |
return $title; |
| 908 |
} |
| 909 |
|
| 910 |
# Check if it's a single post or page |
| 911 |
if ((is_single() || is_page()) && in_the_loop() && is_main_query()) { |
| 912 |
|
| 913 |
# Check if the post was created with the "metasync" system |
| 914 |
$metasync_post = get_post_meta($id, 'metasync_post', true); |
| 915 |
if ($metasync_post === 'yes') { |
| 916 |
return ''; |
| 917 |
} |
| 918 |
} |
| 919 |
return $title; |
| 920 |
} |
| 921 |
|
| 922 |
/** |
| 923 |
* Drop serving/appearance directives when the page is set to noindex. |
| 924 |
* |
| 925 |
* max-snippet, max-image-preview, max-video-preview, nosnippet, noarchive |
| 926 |
* and noimageindex only affect how an indexed page is displayed, so they |
| 927 |
* are meaningless once the page is excluded from the index. When noindex is |
| 928 |
* present we keep only the indexing/crawling directives (noindex, nofollow) |
| 929 |
* so the emitted tag stays clean, e.g. "noindex, nofollow" rather than |
| 930 |
* "noindex, nofollow, noarchive, nosnippet, max-snippet:100, ...". |
| 931 |
* |
| 932 |
* @param string[] $directives |
| 933 |
* @return string[] |
| 934 |
*/ |
| 935 |
private function strip_serving_directives_if_noindex($directives) { |
| 936 |
$has_noindex = false; |
| 937 |
foreach ($directives as $d) { |
| 938 |
if (strcasecmp((string) $d, 'noindex') === 0) { |
| 939 |
$has_noindex = true; |
| 940 |
break; |
| 941 |
} |
| 942 |
} |
| 943 |
if (!$has_noindex) { |
| 944 |
return $directives; |
| 945 |
} |
| 946 |
return array_values(array_filter($directives, function ($d) { |
| 947 |
$d = strtolower((string) $d); |
| 948 |
return $d === 'noindex' || $d === 'nofollow'; |
| 949 |
})); |
| 950 |
} |
| 951 |
|
| 952 |
/** |
| 953 |
* Resolve the robots meta value, honouring the two robots feature switches. |
| 954 |
* |
| 955 |
* Common and advanced directives share one `<meta name="robots">` tag, and |
| 956 |
* they can arrive together from any single storage format — a REST |
| 957 |
* `meta_robots` string may hold "noindex, max-snippet:50". So rather than |
| 958 |
* gate each storage branch, the full value is resolved first and then the |
| 959 |
* directives belonging to a disabled half are dropped. That way every |
| 960 |
* storage path is covered, including the ones that return early. |
| 961 |
* |
| 962 |
* @param array $all_meta Result of get_post_meta($id) (all meta as arrays). |
| 963 |
* @return string Robots directive string, or empty. |
| 964 |
*/ |
| 965 |
private function resolve_robots_value($all_meta) { |
| 966 |
// Neither half is enabled: contribute nothing at all, so core's own |
| 967 |
// robots tag survives untouched. |
| 968 |
if (Metasync_Feature_Flags::robots_fully_disabled()) { |
| 969 |
return ''; |
| 970 |
} |
| 971 |
|
| 972 |
return $this->filter_robots_by_enabled_features( |
| 973 |
$this->resolve_robots_value_raw($all_meta), |
| 974 |
$all_meta |
| 975 |
); |
| 976 |
} |
| 977 |
|
| 978 |
/** |
| 979 |
* Drops directives owned by a disabled robots feature. |
| 980 |
* |
| 981 |
* The three max-* directives are the "advanced" half; everything else |
| 982 |
* (noindex, nofollow, noarchive, nosnippet, noimageindex) is "common". |
| 983 |
* |
| 984 |
* The raw resolver already dropped serving directives when the post carries |
| 985 |
* noindex. If the common half is then switched off that noindex disappears |
| 986 |
* too, so the strip no longer has anything to justify it — the advanced |
| 987 |
* directives it removed belong to a half the user left enabled. They are |
| 988 |
* therefore re-derived from the raw meta once the surviving directives are |
| 989 |
* known. |
| 990 |
* |
| 991 |
* @param string $value Fully resolved robots string. |
| 992 |
* @param array $all_meta Raw meta, used to recover stripped directives. |
| 993 |
* @return string Filtered robots string, or empty when nothing survives. |
| 994 |
*/ |
| 995 |
private function filter_robots_by_enabled_features($value, $all_meta = []) { |
| 996 |
$value = (string) $value; |
| 997 |
if ($value === '') { |
| 998 |
return ''; |
| 999 |
} |
| 1000 |
|
| 1001 |
$common_enabled = Metasync_Feature_Flags::is_enabled(Metasync_Feature_Flags::COMMON_ROBOTS); |
| 1002 |
$advanced_enabled = Metasync_Feature_Flags::is_enabled(Metasync_Feature_Flags::ADVANCE_ROBOTS); |
| 1003 |
if ($common_enabled && $advanced_enabled) { |
| 1004 |
return $value; |
| 1005 |
} |
| 1006 |
|
| 1007 |
$kept = []; |
| 1008 |
foreach (explode(',', $value) as $directive) { |
| 1009 |
$directive = trim($directive); |
| 1010 |
if ($directive === '') { |
| 1011 |
continue; |
| 1012 |
} |
| 1013 |
$is_advanced = stripos($directive, 'max-snippet') === 0 |
| 1014 |
|| stripos($directive, 'max-image-preview') === 0 |
| 1015 |
|| stripos($directive, 'max-video-preview') === 0; |
| 1016 |
|
| 1017 |
if ($is_advanced ? $advanced_enabled : $common_enabled) { |
| 1018 |
$kept[] = $directive; |
| 1019 |
} |
| 1020 |
} |
| 1021 |
|
| 1022 |
// Advanced still on, but the noindex that suppressed its serving |
| 1023 |
// directives has just been filtered out with the common half. Re-resolve |
| 1024 |
// them from the raw meta, since nothing on the page justifies the strip |
| 1025 |
// any more. Reaching here with an empty $kept means the common half is |
| 1026 |
// off (both-enabled returned early above), so this is exactly the |
| 1027 |
// stripped case and not a genuinely empty robots value. |
| 1028 |
if ($advanced_enabled && empty($kept) && !empty($all_meta)) { |
| 1029 |
foreach (explode(',', $this->resolve_advanced_directives($all_meta)) as $directive) { |
| 1030 |
$directive = trim($directive); |
| 1031 |
if ($directive !== '') { |
| 1032 |
$kept[] = $directive; |
| 1033 |
} |
| 1034 |
} |
| 1035 |
} |
| 1036 |
|
| 1037 |
return !empty($kept) ? implode(', ', array_unique($kept)) : ''; |
| 1038 |
} |
| 1039 |
|
| 1040 |
/** |
| 1041 |
* Resolves only the advanced (max-*) directives, ignoring noindex. |
| 1042 |
* |
| 1043 |
* Mirrors the advanced branches of resolve_robots_value_raw() without its |
| 1044 |
* serving-directive strip, so a post whose noindex belongs to a disabled |
| 1045 |
* feature keeps the max-* directives the user still has switched on. |
| 1046 |
* |
| 1047 |
* @param array $all_meta Result of get_post_meta($id) (all meta as arrays). |
| 1048 |
* @return string Comma-separated advanced directives, or empty. |
| 1049 |
*/ |
| 1050 |
private function resolve_advanced_directives($all_meta) { |
| 1051 |
$directives = []; |
| 1052 |
|
| 1053 |
$advanced_raw = $all_meta['_metasync_robots_advanced'][0] ?? ''; |
| 1054 |
if (!empty($advanced_raw)) { |
| 1055 |
$advanced = json_decode($advanced_raw, true); |
| 1056 |
if (is_array($advanced) && !empty($advanced)) { |
| 1057 |
if (isset($advanced['max_snippet'])) { |
| 1058 |
$directives[] = 'max-snippet:' . (int) $advanced['max_snippet']; |
| 1059 |
} |
| 1060 |
if (isset($advanced['max_image_preview'])) { |
| 1061 |
$directives[] = 'max-image-preview:' . esc_attr($advanced['max_image_preview']); |
| 1062 |
} |
| 1063 |
if (isset($advanced['max_video_preview'])) { |
| 1064 |
$directives[] = 'max-video-preview:' . (int) $advanced['max_video_preview']; |
| 1065 |
} |
| 1066 |
if (!empty($directives)) { |
| 1067 |
return implode(', ', $directives); |
| 1068 |
} |
| 1069 |
} |
| 1070 |
} |
| 1071 |
|
| 1072 |
$adv_raw = $all_meta['metasync_advance_robots'][0] ?? ''; |
| 1073 |
if (!empty($adv_raw)) { |
| 1074 |
$adv = maybe_unserialize($adv_raw); |
| 1075 |
if (is_array($adv)) { |
| 1076 |
if (!empty($adv['max-snippet']['enable'])) { |
| 1077 |
$length = isset($adv['max-snippet']['length']) ? (int) $adv['max-snippet']['length'] : -1; |
| 1078 |
$directives[] = 'max-snippet:' . $length; |
| 1079 |
} |
| 1080 |
if (!empty($adv['max-video-preview']['enable'])) { |
| 1081 |
$length = isset($adv['max-video-preview']['length']) ? (int) $adv['max-video-preview']['length'] : -1; |
| 1082 |
$directives[] = 'max-video-preview:' . $length; |
| 1083 |
} |
| 1084 |
if (!empty($adv['max-image-preview']['enable'])) { |
| 1085 |
$length = isset($adv['max-image-preview']['length']) ? sanitize_text_field($adv['max-image-preview']['length']) : 'large'; |
| 1086 |
$directives[] = 'max-image-preview:' . $length; |
| 1087 |
} |
| 1088 |
} |
| 1089 |
} |
| 1090 |
|
| 1091 |
return !empty($directives) ? implode(', ', $directives) : ''; |
| 1092 |
} |
| 1093 |
|
| 1094 |
/** |
| 1095 |
* Resolve the robots meta value from all storage formats. |
| 1096 |
* |
| 1097 |
* Priority order: |
| 1098 |
* 1. _metasync_robots_advanced (JSON) |
| 1099 |
* 2. meta_robots (string, e.g. "noindex, nofollow") — set via REST API |
| 1100 |
* 3. metasync_common_robots (array) + metasync_advance_robots (array) — admin checkboxes |
| 1101 |
* 4. Global defaults from metasync_options['advance_robots_meta'] |
| 1102 |
* |
| 1103 |
* The noindex directive is sourced from _metasync_robots_index (separate key) |
| 1104 |
* and merged into the final string regardless of which storage format is used. |
| 1105 |
* |
| 1106 |
* @param array $all_meta Result of get_post_meta($id) (all meta as arrays). |
| 1107 |
* @return string Robots directive string (e.g. "nofollow, noarchive, max-snippet:-1") or empty. |
| 1108 |
*/ |
| 1109 |
private function resolve_robots_value_raw($all_meta) { |
| 1110 |
$directives = []; |
| 1111 |
|
| 1112 |
// Merge noindex from _metasync_robots_index (separate key, not part of advanced JSON) |
| 1113 |
$robots_index = $all_meta['_metasync_robots_index'][0] ?? ''; |
| 1114 |
if ($robots_index === 'noindex') { |
| 1115 |
$directives[] = 'noindex'; |
| 1116 |
} |
| 1117 |
|
| 1118 |
// Also honor noindex from the classic Common Robots meta box, which stores |
| 1119 |
// it in metasync_common_robots. Without seeding it here, a noindex set via |
| 1120 |
// that meta box is lost whenever the advanced-JSON path below short-circuits. |
| 1121 |
if (!in_array('noindex', $directives, true)) { |
| 1122 |
$common_index_raw = $all_meta['metasync_common_robots'][0] ?? ''; |
| 1123 |
if (!empty($common_index_raw)) { |
| 1124 |
$common_index = maybe_unserialize($common_index_raw); |
| 1125 |
if (is_array($common_index) && !empty($common_index['noindex'])) { |
| 1126 |
$directives[] = 'noindex'; |
| 1127 |
} |
| 1128 |
} |
| 1129 |
} |
| 1130 |
|
| 1131 |
// 1. Check _metasync_robots_advanced (JSON) |
| 1132 |
$advanced_raw = $all_meta['_metasync_robots_advanced'][0] ?? ''; |
| 1133 |
if (!empty($advanced_raw)) { |
| 1134 |
$advanced = json_decode($advanced_raw, true); |
| 1135 |
if (is_array($advanced) && !empty($advanced)) { |
| 1136 |
// Boolean directives |
| 1137 |
foreach (['nofollow', 'noarchive', 'nosnippet', 'noimageindex'] as $dir) { |
| 1138 |
if (!empty($advanced[$dir])) { |
| 1139 |
$directives[] = $dir; |
| 1140 |
} |
| 1141 |
} |
| 1142 |
// max-* directives |
| 1143 |
if (isset($advanced['max_snippet'])) { |
| 1144 |
$directives[] = 'max-snippet:' . (int) $advanced['max_snippet']; |
| 1145 |
} |
| 1146 |
if (isset($advanced['max_image_preview'])) { |
| 1147 |
$directives[] = 'max-image-preview:' . esc_attr($advanced['max_image_preview']); |
| 1148 |
} |
| 1149 |
if (isset($advanced['max_video_preview'])) { |
| 1150 |
$directives[] = 'max-video-preview:' . (int) $advanced['max_video_preview']; |
| 1151 |
} |
| 1152 |
$directives = $this->strip_serving_directives_if_noindex($directives); |
| 1153 |
return !empty($directives) ? implode(', ', $directives) : ''; |
| 1154 |
} |
| 1155 |
} |
| 1156 |
|
| 1157 |
// 2. Check meta_robots (REST API / direct string) |
| 1158 |
$meta_robots = $all_meta['meta_robots'][0] ?? ''; |
| 1159 |
if (!empty($meta_robots)) { |
| 1160 |
// Prepend noindex from _metasync_robots_index if not already in the string |
| 1161 |
if (!empty($directives) && stripos($meta_robots, 'noindex') === false) { |
| 1162 |
return implode(', ', $directives) . ', ' . $meta_robots; |
| 1163 |
} |
| 1164 |
return $meta_robots; |
| 1165 |
} |
| 1166 |
|
| 1167 |
// 3. Check metasync_common_robots (admin checkbox array) |
| 1168 |
$raw = $all_meta['metasync_common_robots'][0] ?? ''; |
| 1169 |
if (!empty($raw)) { |
| 1170 |
$common = maybe_unserialize($raw); |
| 1171 |
if (is_array($common) && !empty($common)) { |
| 1172 |
$common_directives = array_values(array_filter($common, function ($v) { |
| 1173 |
return !empty($v); |
| 1174 |
})); |
| 1175 |
// 'index' is the implicit default and must never be emitted as a |
| 1176 |
// directive. Dropping it here also de-conflicts legacy posts saved |
| 1177 |
// with both Index and No Index checked, so noindex always wins. |
| 1178 |
$common_directives = array_values(array_filter($common_directives, function ($v) { |
| 1179 |
return strcasecmp((string) $v, 'index') !== 0; |
| 1180 |
})); |
| 1181 |
$directives = array_merge($directives, $common_directives); |
| 1182 |
} |
| 1183 |
} |
| 1184 |
|
| 1185 |
// Also check metasync_advance_robots (admin per-post advanced checkboxes) |
| 1186 |
// Structure: ['max-snippet' => ['enable' => '1', 'length' => 100], ...] |
| 1187 |
$adv_raw = $all_meta['metasync_advance_robots'][0] ?? ''; |
| 1188 |
if (!empty($adv_raw)) { |
| 1189 |
$adv = maybe_unserialize($adv_raw); |
| 1190 |
if (is_array($adv)) { |
| 1191 |
if (!empty($adv['max-snippet']['enable'])) { |
| 1192 |
$length = isset($adv['max-snippet']['length']) ? (int) $adv['max-snippet']['length'] : -1; |
| 1193 |
$directives[] = 'max-snippet:' . $length; |
| 1194 |
} |
| 1195 |
if (!empty($adv['max-video-preview']['enable'])) { |
| 1196 |
$length = isset($adv['max-video-preview']['length']) ? (int) $adv['max-video-preview']['length'] : -1; |
| 1197 |
$directives[] = 'max-video-preview:' . $length; |
| 1198 |
} |
| 1199 |
if (!empty($adv['max-image-preview']['enable'])) { |
| 1200 |
$length = isset($adv['max-image-preview']['length']) ? sanitize_text_field($adv['max-image-preview']['length']) : 'large'; |
| 1201 |
$directives[] = 'max-image-preview:' . $length; |
| 1202 |
} |
| 1203 |
} |
| 1204 |
} |
| 1205 |
|
| 1206 |
if (!empty($directives)) { |
| 1207 |
$directives = $this->strip_serving_directives_if_noindex($directives); |
| 1208 |
return implode(', ', array_unique($directives)); |
| 1209 |
} |
| 1210 |
|
| 1211 |
// 4. Global defaults from metasync_options['advance_robots_meta'] |
| 1212 |
$global_options = Metasync::get_option('advance_robots_meta'); |
| 1213 |
if (empty($global_options)) { |
| 1214 |
$global_options = Metasync::get_option('advance_robots_mata'); |
| 1215 |
} |
| 1216 |
if (is_array($global_options)) { |
| 1217 |
if (!empty($global_options['max-snippet']['enable'])) { |
| 1218 |
$length = isset($global_options['max-snippet']['length']) ? (int) $global_options['max-snippet']['length'] : -1; |
| 1219 |
$directives[] = 'max-snippet:' . $length; |
| 1220 |
} |
| 1221 |
if (!empty($global_options['max-video-preview']['enable'])) { |
| 1222 |
$length = isset($global_options['max-video-preview']['length']) ? (int) $global_options['max-video-preview']['length'] : -1; |
| 1223 |
$directives[] = 'max-video-preview:' . $length; |
| 1224 |
} |
| 1225 |
if (!empty($global_options['max-image-preview']['enable'])) { |
| 1226 |
$length = isset($global_options['max-image-preview']['length']) ? sanitize_text_field($global_options['max-image-preview']['length']) : 'large'; |
| 1227 |
$directives[] = 'max-image-preview:' . $length; |
| 1228 |
} |
| 1229 |
} |
| 1230 |
|
| 1231 |
// AC-1: max-image-preview:large outputs by default even with no configuration. |
| 1232 |
// If no per-post or global setting included max-image-preview, apply the default. |
| 1233 |
$has_max_image = false; |
| 1234 |
foreach ($directives as $d) { |
| 1235 |
if (strpos($d, 'max-image-preview') !== false) { |
| 1236 |
$has_max_image = true; |
| 1237 |
break; |
| 1238 |
} |
| 1239 |
} |
| 1240 |
if (!$has_max_image) { |
| 1241 |
$directives[] = 'max-image-preview:large'; |
| 1242 |
} |
| 1243 |
|
| 1244 |
return !empty($directives) ? implode(', ', array_unique($directives)) : ''; |
| 1245 |
} |
| 1246 |
|
| 1247 |
/** |
| 1248 |
* Get the correct canonical URL for archive pages. |
| 1249 |
* |
| 1250 |
* Handles: Posts page (is_home), category, tag, custom taxonomy, |
| 1251 |
* author, year/month/day date archives. |
| 1252 |
* |
| 1253 |
* @return string Archive URL or empty string if not determinable. |
| 1254 |
*/ |
| 1255 |
private function get_archive_canonical_url() { |
| 1256 |
if (is_home()) { |
| 1257 |
$blog_page_id = (int) get_option('page_for_posts'); |
| 1258 |
if ($blog_page_id > 0) { |
| 1259 |
// Honour a manual MetaSync canonical override on the Posts page |
| 1260 |
// (validated — corrupted legacy rows are ignored,). |
| 1261 |
$override = Metasync_Canonical_Sanitizer::sanitize( |
| 1262 |
get_post_meta($blog_page_id, '_metasync_canonical_url', true) |
| 1263 |
); |
| 1264 |
if ($override !== '') { |
| 1265 |
return esc_url($override); |
| 1266 |
} |
| 1267 |
return get_permalink($blog_page_id); |
| 1268 |
} |
| 1269 |
return home_url('/'); |
| 1270 |
} |
| 1271 |
|
| 1272 |
if (is_category() || is_tag() || is_tax()) { |
| 1273 |
$term = get_queried_object(); |
| 1274 |
if ($term instanceof WP_Term) { |
| 1275 |
$link = get_term_link($term); |
| 1276 |
return is_wp_error($link) ? '' : $link; |
| 1277 |
} |
| 1278 |
return ''; |
| 1279 |
} |
| 1280 |
|
| 1281 |
if (is_author()) { |
| 1282 |
return get_author_posts_url(get_queried_object_id()); |
| 1283 |
} |
| 1284 |
|
| 1285 |
if (is_day()) { |
| 1286 |
return get_day_link(get_query_var('year'), get_query_var('monthnum'), get_query_var('day')); |
| 1287 |
} |
| 1288 |
|
| 1289 |
if (is_month()) { |
| 1290 |
return get_month_link(get_query_var('year'), get_query_var('monthnum')); |
| 1291 |
} |
| 1292 |
|
| 1293 |
if (is_year()) { |
| 1294 |
return get_year_link(get_query_var('year')); |
| 1295 |
} |
| 1296 |
|
| 1297 |
return ''; |
| 1298 |
} |
| 1299 |
|
| 1300 |
/** |
| 1301 |
* Get the canonical URL for a post |
| 1302 |
*/ |
| 1303 |
private function get_canonical_url($post) { |
| 1304 |
# Check for persisted OTTO canonical URL first, then the Canonical meta |
| 1305 |
# box value (meta_canonical). Both are validated so legacy rows corrupted |
| 1306 |
# to the literal "Array" (or stored as arrays) are never emitted. |
| 1307 |
$metasync_canonical = Metasync_Canonical_Sanitizer::sanitize( |
| 1308 |
get_post_meta($post->ID, '_metasync_canonical_url', true) |
| 1309 |
); |
| 1310 |
if ($metasync_canonical === '') { |
| 1311 |
$metasync_canonical = Metasync_Canonical_Sanitizer::sanitize( |
| 1312 |
get_post_meta($post->ID, 'meta_canonical', true) |
| 1313 |
); |
| 1314 |
} |
| 1315 |
|
| 1316 |
if ($metasync_canonical !== '') { |
| 1317 |
return esc_url($metasync_canonical); |
| 1318 |
} |
| 1319 |
|
| 1320 |
# Try to get the permalink using WordPress function |
| 1321 |
$permalink = get_permalink($post->ID); |
| 1322 |
|
| 1323 |
# If permalink is not available or is the default query URL, try alternative methods |
| 1324 |
if (!$permalink || strpos($permalink, '?p=') !== false || strpos($permalink, '?page_id=') !== false) { |
| 1325 |
# Force WordPress to generate the proper permalink by temporarily setting post status |
| 1326 |
$original_status = $post->post_status; |
| 1327 |
if ($post->post_status === 'auto-draft') { |
| 1328 |
$post->post_status = 'publish'; |
| 1329 |
} |
| 1330 |
|
| 1331 |
# Try get_permalink again with the updated status |
| 1332 |
$permalink = get_permalink($post->ID); |
| 1333 |
|
| 1334 |
# Restore original status |
| 1335 |
$post->post_status = $original_status; |
| 1336 |
} |
| 1337 |
|
| 1338 |
# If still not working, use WordPress core functions to build proper permalink |
| 1339 |
if (!$permalink || strpos($permalink, '?p=') !== false || strpos($permalink, '?page_id=') !== false) { |
| 1340 |
# Use WordPress core function that respects permalink structure |
| 1341 |
# This properly handles custom structures, hierarchies, and post types |
| 1342 |
# Load admin function if not already available, Without this it is causing error on post the preview page |
| 1343 |
if (!function_exists('get_sample_permalink')) { |
| 1344 |
require_once ABSPATH . 'wp-admin/includes/post.php'; |
| 1345 |
} |
| 1346 |
$permalink = get_sample_permalink($post->ID); |
| 1347 |
|
| 1348 |
if (is_array($permalink)) { |
| 1349 |
# get_sample_permalink returns array with template and slug |
| 1350 |
# Replace %postname% or %pagename% with actual slug |
| 1351 |
$permalink = str_replace( |
| 1352 |
array('%pagename%', '%postname%'), |
| 1353 |
$post->post_name, |
| 1354 |
$permalink[0] |
| 1355 |
); |
| 1356 |
} |
| 1357 |
|
| 1358 |
# Final fallback: if still problematic, construct URL respecting post type structure |
| 1359 |
if (!$permalink || strpos($permalink, '?p=') !== false || strpos($permalink, '?page_id=') !== false) { |
| 1360 |
if (!empty($post->post_name)) { |
| 1361 |
# For pages, check if there's a parent hierarchy |
| 1362 |
if ($post->post_type === 'page' && $post->post_parent) { |
| 1363 |
# Get parent page path for proper hierarchy |
| 1364 |
$parent = get_post($post->post_parent); |
| 1365 |
$parent_path = ''; |
| 1366 |
|
| 1367 |
# Build full path including all parent pages |
| 1368 |
while ($parent) { |
| 1369 |
$parent_path = $parent->post_name . '/' . $parent_path; |
| 1370 |
$parent = $parent->post_parent ? get_post($parent->post_parent) : null; |
| 1371 |
} |
| 1372 |
|
| 1373 |
$permalink = home_url('/' . $parent_path . $post->post_name . '/'); |
| 1374 |
} else { |
| 1375 |
# For posts and pages without parents, use post type archive base |
| 1376 |
$post_type_obj = get_post_type_object($post->post_type); |
| 1377 |
$slug = $post_type_obj->rewrite['slug'] ?? ''; |
| 1378 |
|
| 1379 |
if ($slug && $post->post_type !== 'page') { |
| 1380 |
$permalink = home_url('/' . $slug . '/' . $post->post_name . '/'); |
| 1381 |
} else { |
| 1382 |
$permalink = home_url('/' . $post->post_name . '/'); |
| 1383 |
} |
| 1384 |
} |
| 1385 |
} else { |
| 1386 |
# Fallback to post ID format if no slug available |
| 1387 |
$permalink = home_url('/?p=' . $post->ID); |
| 1388 |
} |
| 1389 |
} |
| 1390 |
} |
| 1391 |
|
| 1392 |
return $permalink; |
| 1393 |
} |
| 1394 |
|
| 1395 |
/** |
| 1396 |
* Set up indexation controls for archive pages |
| 1397 |
* |
| 1398 |
* Called via template_redirect hook to set up early before any output. |
| 1399 |
* This ensures we can capture and clean robots tags from other plugins. |
| 1400 |
* |
| 1401 |
* Logic: |
| 1402 |
* 1. If user wants to add noindex - always do it (override other plugins) |
| 1403 |
* 2. If user wants to allow indexing - only remove other plugins' tags if override setting is enabled |
| 1404 |
* |
| 1405 |
* @since 1.0.0 |
| 1406 |
*/ |
| 1407 |
public function inject_archive_seo_controls() { |
| 1408 |
// Check if we're on a managed archive type |
| 1409 |
if (!$this->is_managed_archive()) { |
| 1410 |
return; |
| 1411 |
} |
| 1412 |
|
| 1413 |
// Get settings |
| 1414 |
$seo_controls = Metasync::get_option('seo_controls', array()); |
| 1415 |
$should_noindex = $this->should_noindex_archive(); |
| 1416 |
$override_enabled = ($seo_controls['override_robots_tags'] ?? 'false') === 'true' || ($seo_controls['override_robots_tags'] ?? false) === true; |
| 1417 |
|
| 1418 |
// Run buffer if either: |
| 1419 |
// 1. User wants to add noindex (always override other plugins), OR |
| 1420 |
// 2. User wants to allow indexing AND override setting is enabled |
| 1421 |
if ($should_noindex || $override_enabled) { |
| 1422 |
// Override WordPress core robots tag |
| 1423 |
add_filter('wp_robots', array($this, 'override_wp_robots'), 999); |
| 1424 |
|
| 1425 |
// Start buffering to remove other plugins' robots tags |
| 1426 |
add_action('wp_head', array($this, 'start_robots_buffer'), 0); |
| 1427 |
add_action('wp_head', array($this, 'end_robots_buffer'), PHP_INT_MAX); |
| 1428 |
} |
| 1429 |
// Otherwise: User wants to allow indexing AND override is disabled - don't interfere |
| 1430 |
} |
| 1431 |
|
| 1432 |
/** |
| 1433 |
* Check if current page is a managed archive type |
| 1434 |
* |
| 1435 |
* Returns true if we're on any archive type that has indexation controls, |
| 1436 |
* regardless of whether noindex is enabled or not. |
| 1437 |
* |
| 1438 |
* @since 1.0.0 |
| 1439 |
* @return bool True if on a managed archive type, false otherwise |
| 1440 |
*/ |
| 1441 |
private function is_managed_archive() { |
| 1442 |
return is_date() || is_tag() || is_author() || is_category() || is_tax('post_format'); |
| 1443 |
} |
| 1444 |
|
| 1445 |
/** |
| 1446 |
* Check if current archive is empty (has no posts) |
| 1447 |
* |
| 1448 |
* Applies to category, tag, author, and post format archives. |
| 1449 |
* Date archives are excluded as they're typically either indexed |
| 1450 |
* or not entirely (less meaningful to check for empty state). |
| 1451 |
* |
| 1452 |
* @since 1.0.0 |
| 1453 |
* @return bool True if archive is empty (0 posts) |
| 1454 |
*/ |
| 1455 |
private function is_empty_archive() { |
| 1456 |
global $wp_query; |
| 1457 |
|
| 1458 |
// Check taxonomy and author archives (excluding date archives) |
| 1459 |
if (!is_category() && !is_tag() && !is_author() && !is_tax('post_format')) { |
| 1460 |
return false; |
| 1461 |
} |
| 1462 |
|
| 1463 |
// Check if the archive has no posts |
| 1464 |
return $wp_query->post_count === 0; |
| 1465 |
} |
| 1466 |
|
| 1467 |
/** |
| 1468 |
* Check if current archive should be noindexed |
| 1469 |
* |
| 1470 |
* @since 1.0.0 |
| 1471 |
* @return bool True if should add noindex, false otherwise |
| 1472 |
*/ |
| 1473 |
private function should_noindex_archive() { |
| 1474 |
// Get indexation control settings |
| 1475 |
$seo_controls = Metasync::get_option('seo_controls', array()); |
| 1476 |
|
| 1477 |
// Check empty archives setting first (applies to categories/tags only) |
| 1478 |
$noindex_empty_archives = $seo_controls['noindex_empty_archives'] ?? false; |
| 1479 |
if (($noindex_empty_archives === 'true' || $noindex_empty_archives === true) && $this->is_empty_archive()) { |
| 1480 |
return true; |
| 1481 |
} |
| 1482 |
|
| 1483 |
// Date Archives |
| 1484 |
if (is_date()) { |
| 1485 |
$index_date_archives = $seo_controls['index_date_archives'] ?? false; |
| 1486 |
// Handle both string and boolean values for compatibility |
| 1487 |
if ($index_date_archives === 'true' || $index_date_archives === true) { |
| 1488 |
return true; |
| 1489 |
} |
| 1490 |
} |
| 1491 |
|
| 1492 |
// Tag Archives |
| 1493 |
elseif (is_tag()) { |
| 1494 |
$index_tag_archives = $seo_controls['index_tag_archives'] ?? false; |
| 1495 |
// Handle both string and boolean values for compatibility |
| 1496 |
if ($index_tag_archives === 'true' || $index_tag_archives === true) { |
| 1497 |
return true; |
| 1498 |
} |
| 1499 |
} |
| 1500 |
|
| 1501 |
// Author Archives |
| 1502 |
elseif (is_author()) { |
| 1503 |
$index_author_archives = $seo_controls['index_author_archives'] ?? false; |
| 1504 |
// Handle both string and boolean values for compatibility |
| 1505 |
if ($index_author_archives === 'true' || $index_author_archives === true) { |
| 1506 |
return true; |
| 1507 |
} |
| 1508 |
} |
| 1509 |
|
| 1510 |
// Category Archives |
| 1511 |
elseif (is_category()) { |
| 1512 |
$index_category_archives = $seo_controls['index_category_archives'] ?? false; |
| 1513 |
// Handle both string and boolean values for compatibility |
| 1514 |
if ($index_category_archives === 'true' || $index_category_archives === true) { |
| 1515 |
return true; |
| 1516 |
} |
| 1517 |
} |
| 1518 |
|
| 1519 |
// Format Archives (post format taxonomy) |
| 1520 |
elseif (is_tax('post_format')) { |
| 1521 |
$index_format_archives = $seo_controls['index_format_archives'] ?? false; |
| 1522 |
if ($index_format_archives === 'true' || $index_format_archives === true) { |
| 1523 |
return true; |
| 1524 |
} |
| 1525 |
} |
| 1526 |
|
| 1527 |
return false; |
| 1528 |
} |
| 1529 |
|
| 1530 |
/** |
| 1531 |
* Override WordPress core robots directives |
| 1532 |
* |
| 1533 |
* Handles both cases: |
| 1534 |
* 1. When noindex is enabled: Add noindex directive |
| 1535 |
* 2. When indexing is allowed: Return empty array (removes WP core robots tag) |
| 1536 |
* |
| 1537 |
* @since 1.0.0 |
| 1538 |
* @param array $robots Associative array of robots directives |
| 1539 |
* @return array Modified robots directives |
| 1540 |
*/ |
| 1541 |
public function override_wp_robots($robots) { |
| 1542 |
if ($this->should_noindex_archive()) { |
| 1543 |
// Case 1: Add noindex directive |
| 1544 |
return array( |
| 1545 |
'noindex' => true, |
| 1546 |
'follow' => true, |
| 1547 |
'max-image-preview' => 'large', // Maintain good image preview for social sharing |
| 1548 |
); |
| 1549 |
} else { |
| 1550 |
// Case 2: Allow indexing - return empty array to remove WP core robots tag |
| 1551 |
// Default behavior without robots tag is to index |
| 1552 |
return array(); |
| 1553 |
} |
| 1554 |
} |
| 1555 |
|
| 1556 |
/** |
| 1557 |
* Start output buffering to capture robots meta tags from other plugins |
| 1558 |
* |
| 1559 |
* @since 1.0.0 |
| 1560 |
*/ |
| 1561 |
public function start_robots_buffer() { |
| 1562 |
ob_start(); |
| 1563 |
} |
| 1564 |
|
| 1565 |
/** |
| 1566 |
* End output buffering, remove existing robots tags, and optionally add ours |
| 1567 |
* |
| 1568 |
* This handles BOTH cases: |
| 1569 |
* 1. When noindex is enabled: Remove other tags and add noindex tag |
| 1570 |
* 2. When indexing is allowed: Remove other plugins' noindex tags |
| 1571 |
* |
| 1572 |
* This ensures our indexation settings always take precedence over other plugins. |
| 1573 |
* |
| 1574 |
* @since 1.0.0 |
| 1575 |
*/ |
| 1576 |
public function end_robots_buffer() { |
| 1577 |
// Only proceed if we're on a managed archive |
| 1578 |
if (!$this->is_managed_archive()) { |
| 1579 |
ob_end_flush(); |
| 1580 |
return; |
| 1581 |
} |
| 1582 |
|
| 1583 |
// Get buffered content |
| 1584 |
$content = ob_get_clean(); |
| 1585 |
|
| 1586 |
// Remove all existing robots meta tags (handles various formats) |
| 1587 |
// This is done for BOTH noindex and index cases to ensure clean slate |
| 1588 |
|
| 1589 |
// Pattern 1: name="robots" content="..." |
| 1590 |
$content = preg_replace( |
| 1591 |
'/<meta\s+name=["\']robots["\']\s+content=["\'][^"\']*["\']\s*\/?>\s*/i', |
| 1592 |
'', |
| 1593 |
$content |
| 1594 |
); |
| 1595 |
|
| 1596 |
// Pattern 2: content="..." name="robots" |
| 1597 |
$content = preg_replace( |
| 1598 |
'/<meta\s+content=["\'][^"\']*["\']\s+name=["\']robots["\']\s*\/?>\s*/i', |
| 1599 |
'', |
| 1600 |
$content |
| 1601 |
); |
| 1602 |
|
| 1603 |
// Pattern 3: Single quotes or no quotes (rare but possible) |
| 1604 |
$content = preg_replace( |
| 1605 |
"/<meta\s+name='robots'\s+content='[^']*'\s*\/?>\s*/i", |
| 1606 |
'', |
| 1607 |
$content |
| 1608 |
); |
| 1609 |
|
| 1610 |
// Pattern 4: property="robots" (some plugins use property attribute) |
| 1611 |
$content = preg_replace( |
| 1612 |
'/<meta\s+property=["\']robots["\']\s+content=["\'][^"\']*["\']\s*\/?>\s*/i', |
| 1613 |
'', |
| 1614 |
$content |
| 1615 |
); |
| 1616 |
|
| 1617 |
// Pattern 5: Search engine specific directives (googlebot, bingbot, etc.) |
| 1618 |
$content = preg_replace( |
| 1619 |
'/<meta\s+name=["\'](?:googlebot|bingbot|googlebot-news|slurp)["\']\s+content=["\'][^"\']*["\']\s*\/?>\s*/i', |
| 1620 |
'', |
| 1621 |
$content |
| 1622 |
); |
| 1623 |
|
| 1624 |
// Output cleaned content |
| 1625 |
echo $content; |
| 1626 |
|
| 1627 |
// Add our robots tag or comment based on settings |
| 1628 |
if ($this->should_noindex_archive()) { |
| 1629 |
// Case 1: User wants to disallow indexing - add noindex tag |
| 1630 |
echo '<!-- MetaSync Indexation Control: Noindex Applied (Overriding other plugins) -->' . "\n"; |
| 1631 |
echo '<meta name="robots" content="noindex, follow">' . "\n"; |
| 1632 |
} else { |
| 1633 |
// Case 2: User wants to allow indexing AND override is enabled |
| 1634 |
// Just remove other plugins' tags, don't add our own |
| 1635 |
// Default behavior without robots tag is to allow indexing |
| 1636 |
echo '<!-- MetaSync Indexation Control: Index Allowed (Override enabled - Other noindex tags removed) -->' . "\n"; |
| 1637 |
} |
| 1638 |
} |
| 1639 |
|
| 1640 |
/** |
| 1641 |
* Filter taxonomy sitemap entries to exclude disabled archive types |
| 1642 |
* |
| 1643 |
* @since 1.0.0 |
| 1644 |
* @param array $taxonomies Array of taxonomy objects |
| 1645 |
* @return array Modified array of taxonomy objects |
| 1646 |
*/ |
| 1647 |
public function filter_sitemap_taxonomies($taxonomies) { |
| 1648 |
// Get indexation control settings |
| 1649 |
$seo_controls = Metasync::get_option('seo_controls', array()); |
| 1650 |
|
| 1651 |
// Check if tag archives are disabled |
| 1652 |
$index_tag_archives = $seo_controls['index_tag_archives'] ?? false; |
| 1653 |
if ($index_tag_archives === 'true' || $index_tag_archives === true) { |
| 1654 |
// Remove post_tag taxonomy from sitemap |
| 1655 |
unset($taxonomies['post_tag']); |
| 1656 |
} |
| 1657 |
|
| 1658 |
// Check if category archives are disabled |
| 1659 |
$index_category_archives = $seo_controls['index_category_archives'] ?? false; |
| 1660 |
if ($index_category_archives === 'true' || $index_category_archives === true) { |
| 1661 |
// Remove category taxonomy from sitemap |
| 1662 |
unset($taxonomies['category']); |
| 1663 |
} |
| 1664 |
|
| 1665 |
// Check if format archives are disabled |
| 1666 |
$index_format_archives = $seo_controls['index_format_archives'] ?? false; |
| 1667 |
if ($index_format_archives === 'true' || $index_format_archives === true) { |
| 1668 |
// Remove post_format taxonomy from sitemap |
| 1669 |
unset($taxonomies['post_format']); |
| 1670 |
} |
| 1671 |
|
| 1672 |
return $taxonomies; |
| 1673 |
} |
| 1674 |
|
| 1675 |
/** |
| 1676 |
* Filter user sitemap entries to exclude disabled author archives |
| 1677 |
* |
| 1678 |
* @since 1.0.0 |
| 1679 |
* @param array $entry Sitemap entry for user |
| 1680 |
* @param WP_User $user User object |
| 1681 |
* @return array|false Modified sitemap entry or false to exclude |
| 1682 |
*/ |
| 1683 |
public function filter_sitemap_users($entry, $user) { |
| 1684 |
// Get indexation control settings |
| 1685 |
$seo_controls = Metasync::get_option('seo_controls', array()); |
| 1686 |
|
| 1687 |
// Check if author archives are disabled |
| 1688 |
$index_author_archives = $seo_controls['index_author_archives'] ?? false; |
| 1689 |
if ($index_author_archives === 'true' || $index_author_archives === true) { |
| 1690 |
// Exclude this user from sitemap |
| 1691 |
return false; |
| 1692 |
} |
| 1693 |
|
| 1694 |
return $entry; |
| 1695 |
} |
| 1696 |
|
| 1697 |
/** |
| 1698 |
* Filter sitemap providers to exclude disabled archive types |
| 1699 |
* |
| 1700 |
* @since 1.0.0 |
| 1701 |
* @param bool $provider Whether to add the provider |
| 1702 |
* @param string $name Provider name |
| 1703 |
* @return bool Whether to add the provider |
| 1704 |
*/ |
| 1705 |
public function filter_sitemap_providers($provider, $name) { |
| 1706 |
// Get indexation control settings |
| 1707 |
$seo_controls = Metasync::get_option('seo_controls', array()); |
| 1708 |
|
| 1709 |
// Check if users/authors sitemap provider should be disabled |
| 1710 |
if ($name === 'users') { |
| 1711 |
$index_author_archives = $seo_controls['index_author_archives'] ?? false; |
| 1712 |
if ($index_author_archives === 'true' || $index_author_archives === true) { |
| 1713 |
return false; // Exclude users sitemap provider entirely |
| 1714 |
} |
| 1715 |
} |
| 1716 |
|
| 1717 |
return $provider; |
| 1718 |
} |
| 1719 |
|
| 1720 |
/** |
| 1721 |
* Filter sitemap index entries to exclude disabled archive types |
| 1722 |
* |
| 1723 |
* @since 1.0.0 |
| 1724 |
* @param array $sitemap_entry Sitemap entry array |
| 1725 |
* @param string $object_type Object type (posts, taxonomies, users) |
| 1726 |
* @param string $subtype Subtype (post type or taxonomy name) |
| 1727 |
* @param int $page Page number |
| 1728 |
* @return array|false Modified sitemap entry or false to exclude |
| 1729 |
*/ |
| 1730 |
public function filter_sitemap_index_entries($sitemap_entry, $object_type, $subtype, $page) { |
| 1731 |
// Get indexation control settings |
| 1732 |
$seo_controls = Metasync::get_option('seo_controls', array()); |
| 1733 |
|
| 1734 |
// Handle taxonomy sitemaps |
| 1735 |
if ($object_type === 'taxonomies') { |
| 1736 |
// Check if tag archives are disabled |
| 1737 |
if ($subtype === 'post_tag') { |
| 1738 |
$index_tag_archives = $seo_controls['index_tag_archives'] ?? false; |
| 1739 |
if ($index_tag_archives === 'true' || $index_tag_archives === true) { |
| 1740 |
return false; // Exclude from sitemap index |
| 1741 |
} |
| 1742 |
} |
| 1743 |
|
| 1744 |
// Check if category archives are disabled |
| 1745 |
if ($subtype === 'category') { |
| 1746 |
$index_category_archives = $seo_controls['index_category_archives'] ?? false; |
| 1747 |
if ($index_category_archives === 'true' || $index_category_archives === true) { |
| 1748 |
return false; // Exclude from sitemap index |
| 1749 |
} |
| 1750 |
} |
| 1751 |
|
| 1752 |
// Check if format archives are disabled |
| 1753 |
if ($subtype === 'post_format') { |
| 1754 |
$index_format_archives = $seo_controls['index_format_archives'] ?? false; |
| 1755 |
if ($index_format_archives === 'true' || $index_format_archives === true) { |
| 1756 |
return false; // Exclude from sitemap index |
| 1757 |
} |
| 1758 |
} |
| 1759 |
} |
| 1760 |
|
| 1761 |
// Handle user/author sitemaps |
| 1762 |
if ($object_type === 'users') { |
| 1763 |
$index_author_archives = $seo_controls['index_author_archives'] ?? false; |
| 1764 |
if ($index_author_archives === 'true' || $index_author_archives === true) { |
| 1765 |
return false; // Exclude from sitemap index |
| 1766 |
} |
| 1767 |
} |
| 1768 |
|
| 1769 |
return $sitemap_entry; |
| 1770 |
} |
| 1771 |
} |
| 1772 |
|
