| @@ -785,9 +785,9 @@ | ||
| 785 | 785 | $response = wp_remote_get($url_with_params, $args); |
| 786 | 786 | |
| 787 | 787 | # Check for errors |
| 788 | 788 | if (is_wp_error($response)) { |
| 789 | - error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': API call failed - ' . $response->get_error_message()); | |
| 789 | + error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': API call failed - ' . $response->get_error_message()); // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log -- unreachable in production | |
| 790 | 790 | return false; |
| 791 | 791 | } |
| 792 | 792 | |
| 793 | 793 | # get the response body |
| @@ -797,9 +797,9 @@ | ||
| 797 | 797 | $response_code = wp_remote_retrieve_response_code($response); |
| 798 | 798 | |
| 799 | 799 | # if no change data skip |
| 800 | 800 | if (empty($body) || $response_code !== 200){ |
| 801 | - error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': API returned empty or non-200. Code: ' . $response_code); | |
| 801 | + error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': API returned empty or non-200. Code: ' . $response_code); // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log -- unreachable in production | |
| 802 | 802 | return false; |
| 803 | 803 | } |
| 804 | 804 | |
| 805 | 805 | # set the html file path |
| @@ -992,9 +992,11 @@ | ||
| 992 | 992 | |
| 993 | 993 | # Check for timeout or connection errors |
| 994 | 994 | if (is_wp_error($route_html)) { |
| 995 | 995 | $error_msg = $route_html->get_error_message(); |
| 996 | - error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ' DEBUG: FAILED - wp_remote_get error: ' . $error_msg . ' for route: ' . $route); | |
| 996 | + if (function_exists('metasync_otto_report_render_failure')) { | |
| 997 | + metasync_otto_report_render_failure('HTTP_FETCH_FAILED', 'Internal fetch failed: ' . $error_msg, [], 'warning'); | |
| 998 | + } | |
| 997 | 999 | return false; |
| 998 | 1000 | } |
| 999 | 1001 | |
| 1000 | 1002 | # get body |
| @@ -1002,8 +1004,18 @@ | ||
| 1002 | 1004 | |
| 1003 | 1005 | # Get the response code |
| 1004 | 1006 | $response_code = wp_remote_retrieve_response_code($route_html); |
| 1005 | 1007 | |
| 1008 | + // Requests follows redirects and returns the final body with its final | |
| 1009 | + // status. Do not put that body on the original route when a redirect | |
| 1010 | + // changed the URL: doing so creates a soft redirect with mismatched | |
| 1011 | + // content and SEO metadata. The normal WordPress request can handle | |
| 1012 | + // the redirect, and OTTO will process the destination on its own URL. | |
| 1013 | + $effective_url = $this->get_effective_fetch_url($route_html); | |
| 1014 | + if ($effective_url !== '' && !$this->fetch_url_matches_route($effective_url, $route, $request_body)) { | |
| 1015 | + return false; | |
| 1016 | + } | |
| 1017 | + | |
| 1006 | 1018 | # Capture the internal fetch's `Set-Cookie`. On the HTTP path OTTO keeps only the |
| 1007 | 1019 | # body below and would otherwise discard this — but it is both the form/session |
| 1008 | 1020 | # signal for the caching decision AND (same-origin only) the session the visitor |
| 1009 | 1021 | # needs for their forms to work. $request_body is the URL actually fetched (it may |
| @@ -1032,9 +1044,11 @@ | ||
| 1032 | 1044 | |
| 1033 | 1045 | |
| 1034 | 1046 | # check not empty |
| 1035 | 1047 | if(empty($html_body) || $response_code !== 200){ |
| 1036 | - error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ' DEBUG: FAILED - Empty body or non-200 status for route: ' . $route); | |
| 1048 | + if (function_exists('metasync_otto_report_render_failure')) { | |
| 1049 | + metasync_otto_report_render_failure('HTTP_FETCH_EMPTY', 'Internal fetch returned empty body or non-200 status (HTTP ' . intval($response_code) . ')', [], 'warning'); | |
| 1050 | + } | |
| 1037 | 1051 | return false; |
| 1038 | 1052 | } |
| 1039 | 1053 | |
| 1040 | 1054 | # Skip DOM processing on oversized documents to avoid fatal OOM. |
| @@ -1198,9 +1212,9 @@ | ||
| 1198 | 1212 | |
| 1199 | 1213 | $result_html = preg_replace_callback( |
| 1200 | 1214 | '/(<' . $heading_type . '(?:\s[^>]*)?>)(.*?)(<\/' . $heading_type . '>)/is', |
| 1201 | 1215 | function ($m) use ($current_value, $recommended_value) { |
| 1202 | - $inner_text = trim(preg_replace('/\s+/', ' ', html_entity_decode(strip_tags($m[2]), ENT_QUOTES, 'UTF-8'))); | |
| 1216 | + $inner_text = trim(preg_replace('/\s+/', ' ', html_entity_decode(wp_strip_all_tags($m[2]), ENT_QUOTES, 'UTF-8'))); | |
| 1203 | 1217 | if ($inner_text !== $current_value) { |
| 1204 | 1218 | return $m[0]; |
| 1205 | 1219 | } |
| 1206 | 1220 | # Keep the author's inline formatting; skip when it can't |
| @@ -1222,18 +1236,16 @@ | ||
| 1222 | 1236 | $result_html = $this->otto_guard_html($this->apply_image_alt_text_via_string($result_html, $replacement_data), $result_html, 'apply_image_alt_text_via_string'); |
| 1223 | 1237 | |
| 1224 | 1238 | # Apply insertions — only if DOM insertion didn't already apply it |
| 1225 | 1239 | if (!empty($replacement_data['header_html_insertion'])) { |
| 1226 | - # Escape JSON-LD payload closers first; the DOM path inserts the | |
| 1227 | - # same escaped bytes, so the already-applied check must match them. | |
| 1228 | - $escaped_header_insertion = metasync_escape_json_ld_blocks_in_html($replacement_data['header_html_insertion']); | |
| 1229 | - $header_html_check = trim($escaped_header_insertion); | |
| 1240 | + # Escape JSON-LD payload closers and stamp OTTO's blocks first; the DOM | |
| 1241 | + # path inserts the same bytes, so the already-applied check must compare | |
| 1242 | + # against the stamped form or it would insert a second copy. | |
| 1243 | + $header_html_insertion = $this->stamp_otto_json_ld( | |
| 1244 | + metasync_escape_json_ld_blocks_in_html($replacement_data['header_html_insertion']) | |
| 1245 | + ); | |
| 1246 | + $header_html_check = trim($header_html_insertion); | |
| 1230 | 1247 | if (strpos($result_html, $header_html_check) === false) { |
| 1231 | - $header_html_insertion = preg_replace( | |
| 1232 | - '/<script(\s[^>]*)type\s*=\s*(["\'])application\/ld\+json\2/i', | |
| 1233 | - '<script$1type=$2application/ld+json$2 data-otto="true"', | |
| 1234 | - $escaped_header_insertion | |
| 1235 | - ); | |
| 1236 | 1248 | $safe_header = str_replace(array('\\', '$'), array('\\\\', '\\$'), $header_html_insertion); |
| 1237 | 1249 | $result_html = preg_replace('/(<\/head>)/i', $safe_header . "\n" . '$1', $result_html, 1); |
| 1238 | 1250 | } |
| 1239 | 1251 | } |
| @@ -1261,8 +1273,9 @@ | ||
| 1261 | 1273 | |
| 1262 | 1274 | # DEDUPLICATION: Remove duplicate <title>, meta description, OG, Twitter tags, canonical, and JSON-LD schema |
| 1263 | 1275 | $result_html = $this->otto_guard_html($this->deduplicate_title_tags($result_html), $result_html, 'deduplicate_title_tags'); |
| 1264 | 1276 | $result_html = $this->otto_guard_html($this->deduplicate_description_tags($result_html), $result_html, 'deduplicate_description_tags'); |
| 1277 | + $result_html = $this->otto_guard_html($this->deduplicate_keywords_tags($result_html), $result_html, 'deduplicate_keywords_tags'); | |
| 1265 | 1278 | $result_html = $this->otto_guard_html($this->deduplicate_og_twitter_tags($result_html), $result_html, 'deduplicate_og_twitter_tags'); |
| 1266 | 1279 | $result_html = $this->otto_guard_html($this->apply_metabox_og_precedence($result_html), $result_html, 'apply_metabox_og_precedence'); |
| 1267 | 1280 | $result_html = $this->otto_guard_html($this->apply_custom_seo_precedence($result_html), $result_html, 'apply_custom_seo_precedence'); |
| 1268 | 1281 | $result_html = $this->otto_guard_html($this->deduplicate_schema_tags($result_html), $result_html, 'deduplicate_schema_tags'); |
| @@ -1323,8 +1336,111 @@ | ||
| 1323 | 1336 | return $result_html; |
| 1324 | 1337 | } |
| 1325 | 1338 | |
| 1326 | 1339 | /** |
| 1340 | + * Read the final URL from the WordPress HTTP response when Requests has | |
| 1341 | + * followed one or more redirects. Older HTTP transports may not expose it; | |
| 1342 | + * in that case the empty result leaves the existing fail-open behavior. | |
| 1343 | + * | |
| 1344 | + * @param mixed $response wp_remote_get() response. | |
| 1345 | + * @return string Final URL, or an empty string when unavailable. | |
| 1346 | + */ | |
| 1347 | + private function get_effective_fetch_url($response) | |
| 1348 | + { | |
| 1349 | + if (!is_array($response) || empty($response['http_response']) || !is_object($response['http_response'])) { | |
| 1350 | + return ''; | |
| 1351 | + } | |
| 1352 | + | |
| 1353 | + $http_response = $response['http_response']; | |
| 1354 | + if (!method_exists($http_response, 'get_response_object')) { | |
| 1355 | + return ''; | |
| 1356 | + } | |
| 1357 | + | |
| 1358 | + $requests_response = $http_response->get_response_object(); | |
| 1359 | + if (!is_object($requests_response) || empty($requests_response->url) || !is_string($requests_response->url)) { | |
| 1360 | + return ''; | |
| 1361 | + } | |
| 1362 | + | |
| 1363 | + return $requests_response->url; | |
| 1364 | + } | |
| 1365 | + | |
| 1366 | + /** | |
| 1367 | + * Determine whether a followed response still represents the requested | |
| 1368 | + * route. The internal fetch may use a rewritten host for tunnel support, | |
| 1369 | + * so both the original and transport hosts are accepted; the path and | |
| 1370 | + * non-OTTO query parameters must remain identical. | |
| 1371 | + * | |
| 1372 | + * @param string $effective_url Final URL reported by Requests. | |
| 1373 | + * @param string $route Original route supplied to OTTO. | |
| 1374 | + * @param string $fetch_url URL actually passed to wp_remote_get(). | |
| 1375 | + * @return bool True when no meaningful redirect changed the route. | |
| 1376 | + */ | |
| 1377 | + private function fetch_url_matches_route($effective_url, $route, $fetch_url) | |
| 1378 | + { | |
| 1379 | + $effective = wp_parse_url($effective_url); | |
| 1380 | + $requested = wp_parse_url($route); | |
| 1381 | + $transport = wp_parse_url($fetch_url); | |
| 1382 | + if (!is_array($effective) || !is_array($requested)) { | |
| 1383 | + return true; | |
| 1384 | + } | |
| 1385 | + | |
| 1386 | + $effective_path = isset($effective['path']) ? $effective['path'] : '/'; | |
| 1387 | + $requested_path = isset($requested['path']) ? $requested['path'] : '/'; | |
| 1388 | + if ($effective_path !== $requested_path) { | |
| 1389 | + return false; | |
| 1390 | + } | |
| 1391 | + | |
| 1392 | + $allowed_hosts = []; | |
| 1393 | + foreach ([$requested, $transport] as $parts) { | |
| 1394 | + if (!empty($parts['host'])) { | |
| 1395 | + $host = strtolower($parts['host']); | |
| 1396 | + if (!empty($parts['port'])) { | |
| 1397 | + $host .= ':' . $parts['port']; | |
| 1398 | + } | |
| 1399 | + $allowed_hosts[] = $host; | |
| 1400 | + } | |
| 1401 | + } | |
| 1402 | + if (!empty($effective['host'])) { | |
| 1403 | + $effective_host = strtolower($effective['host']); | |
| 1404 | + if (!empty($effective['port'])) { | |
| 1405 | + $effective_host .= ':' . $effective['port']; | |
| 1406 | + } | |
| 1407 | + if (!empty($allowed_hosts) && !in_array($effective_host, $allowed_hosts, true)) { | |
| 1408 | + return false; | |
| 1409 | + } | |
| 1410 | + } | |
| 1411 | + | |
| 1412 | + return $this->without_otto_fetch_params($effective['query'] ?? '') | |
| 1413 | + === $this->without_otto_fetch_params($requested['query'] ?? ''); | |
| 1414 | + } | |
| 1415 | + | |
| 1416 | + /** | |
| 1417 | + * Remove parameters used only to mark OTTO's internal request. | |
| 1418 | + * | |
| 1419 | + * @param string $query Query string without the leading '?'. | |
| 1420 | + * @return string Query with OTTO markers removed. | |
| 1421 | + */ | |
| 1422 | + private function without_otto_fetch_params($query) | |
| 1423 | + { | |
| 1424 | + $query = (string) $query; | |
| 1425 | + if ($query === '') { | |
| 1426 | + return ''; | |
| 1427 | + } | |
| 1428 | + | |
| 1429 | + $parts = []; | |
| 1430 | + foreach (explode('&', $query) as $part) { | |
| 1431 | + $name = strstr($part, '=', true); | |
| 1432 | + $name = $name === false ? $part : $name; | |
| 1433 | + if (in_array($name, ['is_otto_page_fetch', 'otto_block_title', 'otto_block_desc'], true)) { | |
| 1434 | + continue; | |
| 1435 | + } | |
| 1436 | + $parts[] = $part; | |
| 1437 | + } | |
| 1438 | + | |
| 1439 | + return implode('&', $parts); | |
| 1440 | + } | |
| 1441 | + | |
| 1442 | + /** | |
| 1327 | 1443 | * Repair double-encoded character references in serialized HTML. |
| 1328 | 1444 | * |
| 1329 | 1445 | * The bundled simplehtmldom serializer (HtmlNode::makeup(), reached via |
| 1330 | 1446 | * $this->dom->save() and $root->outertext) runs every attribute value through |
| @@ -1985,10 +2101,10 @@ | ||
| 1985 | 2101 | } |
| 1986 | 2102 | |
| 1987 | 2103 | # If it's a protocol-relative URL (//example.com), check if it matches our domain |
| 1988 | 2104 | if (strpos($url, '//') === 0) { |
| 1989 | - $parsed_home = parse_url($home_url); | |
| 1990 | - $parsed_link = parse_url($url); | |
| 2105 | + $parsed_home = wp_parse_url($home_url); | |
| 2106 | + $parsed_link = wp_parse_url($url); | |
| 1991 | 2107 | |
| 1992 | 2108 | if (isset($parsed_home['host']) && isset($parsed_link['host'])) { |
| 1993 | 2109 | if (strtolower($parsed_home['host']) === strtolower($parsed_link['host'])) { |
| 1994 | 2110 | return false; |
| @@ -2110,21 +2226,32 @@ | ||
| 2110 | 2226 | return $post_id ? (int) $post_id : 0; |
| 2111 | 2227 | } |
| 2112 | 2228 | |
| 2113 | 2229 | /** |
| 2114 | - * Replace the first <meta $attr="$val"> content with $content_escaped (marked | |
| 2115 | - * data-metasync-seo="custom"); insert one after <head> if none is present. | |
| 2116 | - * Helper for apply_custom_seo_precedence(). | |
| 2230 | + * Replace the first <meta $attr="$val"> content with $content_escaped. | |
| 2117 | 2231 | * |
| 2118 | 2232 | * @param string $html |
| 2119 | 2233 | * @param string $attr "name" or "property" |
| 2120 | 2234 | * @param string $val e.g. "description", "og:description" |
| 2121 | 2235 | * @param string $content_escaped Already HTML-escaped content value |
| 2236 | + * @param string $marker Provenance marker for the winning tier | |
| 2122 | 2237 | * @return string |
| 2123 | 2238 | */ |
| 2124 | - private function force_custom_meta($html, $attr, $val, $content_escaped) { | |
| 2125 | - $tag = '<meta ' . $attr . '="' . $val . '" content="' . $content_escaped . '" data-metasync-seo="custom" />'; | |
| 2126 | - $pattern = '/<meta\s[^>]*' . preg_quote($attr, '/') . '\s*=\s*["\']' . preg_quote($val, '/') . '["\'][^>]*\/?>/i'; | |
| 2239 | + private function force_custom_meta($html, $attr, $val, $content_escaped, $marker = 'custom') { | |
| 2240 | + $marker_attribute = $marker === 'otto' | |
| 2241 | + ? ' data-metasync-otto="true"' | |
| 2242 | + : ' data-metasync-seo="custom"'; | |
| 2243 | + $tag = '<meta ' . $attr . '="' . $val . '" content="' . $content_escaped . '"' . $marker_attribute . ' />'; | |
| 2244 | + # The optional prefix group must end on whitespace or a quote so the | |
| 2245 | + # attribute matched is $attr itself, not something merely ending in it: | |
| 2246 | + # a bare [^>]*name= also matches data-name="keywords", so a third party's | |
| 2247 | + # unrelated tag was consumed and rewritten while the real one survived — | |
| 2248 | + # destroying their value AND leaving the duplicate the dedup had just | |
| 2249 | + # collapsed. Making the group optional keeps the common | |
| 2250 | + # <meta name="..." ...> form matching, where <meta\s has already consumed | |
| 2251 | + # the only separator. Same rule as deduplicate_keywords_tags(); this | |
| 2252 | + # helper serves title, description and keywords, so all three are covered. | |
| 2253 | + $pattern = '/<meta\s(?:[^>]*[\s"\'])?' . preg_quote($attr, '/') . '\s*=\s*["\']' . preg_quote($val, '/') . '["\'][^>]*\/?>/i'; | |
| 2127 | 2254 | $count = 0; |
| 2128 | 2255 | $new = preg_replace_callback($pattern, function ($m) use ($tag) { |
| 2129 | 2256 | return $tag; |
| 2130 | 2257 | }, $html, 1, $count); |
| @@ -2137,23 +2264,37 @@ | ||
| 2137 | 2264 | }, $html, 1); |
| 2138 | 2265 | } |
| 2139 | 2266 | |
| 2140 | 2267 | /** |
| 2141 | - * Enforce the SEO-sidebar custom title/description as the FINAL word over OTTO. | |
| 2268 | + * Make sure the precedence resolver is loaded before calling it. | |
| 2142 | 2269 | * |
| 2143 | - * The DOM-pass guards (do_header_replacements/handle_meta_element) already skip | |
| 2144 | - * OTTO for these fields, but later steps re-introduce OTTO's value: | |
| 2145 | - * - the manual string-replacement title pass re-applies OTTO's title; | |
| 2146 | - * - OTTO's description arrives via header_html_insertion (data-otto) and the | |
| 2147 | - * dedup keeps it whenever no data-metasync-seo tag is present — which is | |
| 2148 | - * exactly the case when the post is synced to a third-party SEO plugin and | |
| 2149 | - * the sidebar deferred emitting its own tag (output_seo_meta_description()). | |
| 2270 | + * @return bool | |
| 2271 | + */ | |
| 2272 | + private static function precedence_available() { | |
| 2273 | + if (class_exists('Metasync_Seo_Precedence')) { | |
| 2274 | + return true; | |
| 2275 | + } | |
| 2276 | + | |
| 2277 | + $file = dirname(__DIR__) . '/includes/class-metasync-seo-precedence.php'; | |
| 2278 | + if (is_readable($file)) { | |
| 2279 | + require_once $file; | |
| 2280 | + } | |
| 2281 | + | |
| 2282 | + return class_exists('Metasync_Seo_Precedence'); | |
| 2283 | + } | |
| 2284 | + | |
| 2285 | + /** | |
| 2286 | + * Enforce the resolved SEO title, description and keywords as the FINAL | |
| 2287 | + * word over OTTO. | |
| 2150 | 2288 | * |
| 2151 | - * Running LAST — after all dedup/precedence steps — this makes the user's | |
| 2152 | - * explicit sidebar value authoritative regardless of how the tag was produced, | |
| 2153 | - * mirroring apply_metabox_og_precedence() for the OG meta box. Singular views | |
| 2154 | - * only; a value must be non-empty (explicitly set) to take precedence. | |
| 2289 | + * The resolver owns the global custom-versus-OTTO order. This final buffer | |
| 2290 | + * pass must use it too: otherwise the setting works through the sidebar | |
| 2291 | + * and third-party filters but OTTO's later string pass puts the custom | |
| 2292 | + * value back on the page. | |
| 2155 | 2293 | * |
| 2294 | + * OG and Twitter descriptions remain outside this method deliberately; | |
| 2295 | + * social precedence is a separate follow-up scope. | |
| 2296 | + * | |
| 2156 | 2297 | * @param string $html Full HTML document (post-dedup). |
| 2157 | 2298 | * @return string |
| 2158 | 2299 | */ |
| 2159 | 2300 | private function apply_custom_seo_precedence($html) { |
| @@ -2164,12 +2305,26 @@ | ||
| 2164 | 2305 | if (!$post_id) { |
| 2165 | 2306 | return $html; |
| 2166 | 2307 | } |
| 2167 | 2308 | |
| 2168 | - # Title — force the custom sidebar title into the (single, post-dedup) <title>. | |
| 2169 | - $custom_title = get_post_meta($post_id, '_metasync_seo_title', true); | |
| 2170 | - if (!empty($custom_title) && is_string($custom_title)) { | |
| 2171 | - $title_tag = '<title>' . htmlspecialchars($custom_title, ENT_QUOTES, 'UTF-8') . '</title>'; | |
| 2309 | + # This runs inside the output buffer, where an unresolvable class is a | |
| 2310 | + # fatal that takes the whole page down — otto_guard_html() only catches | |
| 2311 | + # a NULL return, not a thrown Error. A partial update can leave newer | |
| 2312 | + # PHP files beside an older committed autoload classmap, so resolve the | |
| 2313 | + # file directly before trusting the class, and leave the document | |
| 2314 | + # untouched if it is genuinely absent. Same reasoning as | |
| 2315 | + # Metasync_SEO_Conflict_Handler::precedence_available(). | |
| 2316 | + if (!self::precedence_available()) { | |
| 2317 | + return $html; | |
| 2318 | + } | |
| 2319 | + | |
| 2320 | + # Title — replace the resolved page title into the (single, post-dedup) <title>. | |
| 2321 | + $resolved_title = Metasync_Seo_Precedence::resolve( | |
| 2322 | + $post_id, | |
| 2323 | + Metasync_Seo_Precedence::FIELD_TITLE | |
| 2324 | + ); | |
| 2325 | + if (!empty($resolved_title['value'])) { | |
| 2326 | + $title_tag = '<title>' . htmlspecialchars($resolved_title['value'], ENT_QUOTES, 'UTF-8') . '</title>'; | |
| 2172 | 2327 | $count = 0; |
| 2173 | 2328 | $replaced = preg_replace_callback('/<title[^>]*>.*?<\/title>/is', function ($m) use ($title_tag) { |
| 2174 | 2329 | return $title_tag; |
| 2175 | 2330 | }, $html, 1, $count); |
| @@ -2181,19 +2336,59 @@ | ||
| 2181 | 2336 | }, $html, 1); |
| 2182 | 2337 | } |
| 2183 | 2338 | } |
| 2184 | 2339 | |
| 2185 | - # Description — force the custom sidebar description into the SEO | |
| 2186 | - # <meta name="description"> only. og:/twitter: description are social tags | |
| 2187 | - # owned by the OG meta box and apply_metabox_og_precedence() (which runs | |
| 2188 | - # earlier); this pass must NOT touch them, or it would override a | |
| 2189 | - # separately user-set og:description with the plain SEO description. | |
| 2190 | - $custom_desc = get_post_meta($post_id, '_metasync_seo_desc', true); | |
| 2191 | - if (!empty($custom_desc) && is_string($custom_desc)) { | |
| 2192 | - $esc = htmlspecialchars($custom_desc, ENT_QUOTES, 'UTF-8'); | |
| 2193 | - $html = $this->force_custom_meta($html, 'name', 'description', $esc); | |
| 2340 | + # Description — replace the resolved page description into the SEO | |
| 2341 | + # <meta name="description"> only. OG/Twitter description is social | |
| 2342 | + # precedence and remains outside this setting's scope. | |
| 2343 | + $resolved_description = Metasync_Seo_Precedence::resolve( | |
| 2344 | + $post_id, | |
| 2345 | + Metasync_Seo_Precedence::FIELD_DESCRIPTION | |
| 2346 | + ); | |
| 2347 | + if (!empty($resolved_description['value'])) { | |
| 2348 | + $esc = htmlspecialchars($resolved_description['value'], ENT_QUOTES, 'UTF-8'); | |
| 2349 | + $marker = Metasync_Seo_Precedence::is_otto_value( | |
| 2350 | + $post_id, | |
| 2351 | + Metasync_Seo_Precedence::FIELD_DESCRIPTION, | |
| 2352 | + $resolved_description | |
| 2353 | + ) | |
| 2354 | + ? 'otto' | |
| 2355 | + : 'custom'; | |
| 2356 | + $html = $this->force_custom_meta($html, 'name', 'description', $esc, $marker); | |
| 2194 | 2357 | } |
| 2195 | 2358 | |
| 2359 | + # Keywords — the same final pass as the description above, on the chain | |
| 2360 | + # the resolver owns, so the global priority setting governs this tag | |
| 2361 | + # too. OTTO's keywords tag arrives via header_html_insertion carrying | |
| 2362 | + # a data-otto marker and is spliced in after wp_head, so it is always | |
| 2363 | + # the later tag and always wins deduplicate_keywords_tags(); forcing | |
| 2364 | + # the resolved winner here is what keeps the losing tier reachable at | |
| 2365 | + # all — the single-printing-owner rule the description branch follows. | |
| 2366 | + # | |
| 2367 | + # The tier this forces is NOT guaranteed customer-typed, unlike the | |
| 2368 | + # title and description keys above: the meta_keywords persistence flag | |
| 2369 | + # writes OTTO's value straight to _metasync_focus_keyword, and neither | |
| 2370 | + # that conditional write nor the clone cleaner (which strips only the | |
| 2371 | + # _metasync_otto_* keys) ever clears it, so a stale flag copy can | |
| 2372 | + # outrank a newer suggestion under custom-first. That key sharing is | |
| 2373 | + # tracked separately — the fix is a key of its own for the persisted | |
| 2374 | + # keyword, not a special case here. | |
| 2375 | + $resolved_keyword = Metasync_Seo_Precedence::resolve( | |
| 2376 | + $post_id, | |
| 2377 | + Metasync_Seo_Precedence::FIELD_FOCUS_KEYWORD | |
| 2378 | + ); | |
| 2379 | + if (!empty($resolved_keyword['value'])) { | |
| 2380 | + $esc = htmlspecialchars($resolved_keyword['value'], ENT_QUOTES, 'UTF-8'); | |
| 2381 | + $marker = Metasync_Seo_Precedence::is_otto_value( | |
| 2382 | + $post_id, | |
| 2383 | + Metasync_Seo_Precedence::FIELD_FOCUS_KEYWORD, | |
| 2384 | + $resolved_keyword | |
| 2385 | + ) | |
| 2386 | + ? 'otto' | |
| 2387 | + : 'custom'; | |
| 2388 | + $html = $this->force_custom_meta($html, 'name', 'keywords', $esc, $marker); | |
| 2389 | + } | |
| 2390 | + | |
| 2196 | 2391 | return $html; |
| 2197 | 2392 | } |
| 2198 | 2393 | |
| 2199 | 2394 | # this function does the header replacements |
| @@ -2938,8 +3133,82 @@ | ||
| 2938 | 3133 | return $html; |
| 2939 | 3134 | } |
| 2940 | 3135 | |
| 2941 | 3136 | /** |
| 3137 | + * Remove duplicate <meta name="keywords"> tags from HTML. | |
| 3138 | + * | |
| 3139 | + * OTTO ships its keywords tag inside header_html_insertion, and MetaSync's | |
| 3140 | + * own SEO output now emits one too whenever the precedence chain resolves a | |
| 3141 | + * keyword — so a server-side-rendered page can carry both. | |
| 3142 | + * | |
| 3143 | + * The keeper is OTTO's tag, or the first occurrence when no OTTO tag is | |
| 3144 | + * present. Unlike deduplicate_description_tags() there is no | |
| 3145 | + * customer-authored branch here: print_metatag() emits no data attribute, so | |
| 3146 | + * a keywords tag never carries data-metasync-seo="custom" at this point in | |
| 3147 | + * the pipeline. The customer tier is re-applied afterwards by | |
| 3148 | + * apply_custom_seo_precedence(), which runs later and stamps the marker via | |
| 3149 | + * force_custom_meta(). Collapsing to OTTO's tag here is therefore safe: it is | |
| 3150 | + * not the final word on what the page renders. | |
| 3151 | + * | |
| 3152 | + * @param string $html Full HTML document. | |
| 3153 | + * @return string HTML with at most one <meta name="keywords">. | |
| 3154 | + */ | |
| 3155 | + private function deduplicate_keywords_tags($html) { | |
| 3156 | + # No is_string() guard, unlike deduplicate_description_tags() beside it. | |
| 3157 | + # That asymmetry is forced, not an oversight: @param string makes the | |
| 3158 | + # check provably always-true at PHPStan level 5, and the description | |
| 3159 | + # sibling's copy is one of a fixed number of occurrences the baseline | |
| 3160 | + # pins by count, so adding another fails the build. Both methods are | |
| 3161 | + # only ever called with the buffer string. | |
| 3162 | + if ($html === '') { | |
| 3163 | + return $html; | |
| 3164 | + } | |
| 3165 | + | |
| 3166 | + # Match <meta ... name="keywords" ...> in either attribute order. | |
| 3167 | + # | |
| 3168 | + # The optional prefix group must end on whitespace or a quote so the | |
| 3169 | + # attribute is `name`, not something merely ending in it: a bare | |
| 3170 | + # [^>]*name= also matches data-name="keywords". Making the group | |
| 3171 | + # optional keeps the common `<meta name="keywords" ...>` form matching, | |
| 3172 | + # where <meta\s has already consumed the only separator. | |
| 3173 | + # | |
| 3174 | + # The opening quote must be immediately followed by "keywords", so | |
| 3175 | + # name="news_keywords" is not matched. | |
| 3176 | + $pattern = '/<meta\s(?:[^>]*[\s"\'])?name\s*=\s*["\']keywords["\'][^>]*\/?>/i'; | |
| 3177 | + | |
| 3178 | + if (preg_match_all($pattern, $html, $matches) <= 1) { | |
| 3179 | + return $html; # 0 or 1 — nothing to deduplicate | |
| 3180 | + } | |
| 3181 | + | |
| 3182 | + $all_tags = $matches[0]; | |
| 3183 | + | |
| 3184 | + $otto_tag = null; | |
| 3185 | + foreach ($all_tags as $tag) { | |
| 3186 | + if (stripos($tag, 'data-otto-pixel') !== false || | |
| 3187 | + stripos($tag, 'data-metasync-otto') !== false || | |
| 3188 | + stripos($tag, 'data-otto') !== false | |
| 3189 | + ) { | |
| 3190 | + $otto_tag = $tag; | |
| 3191 | + break; | |
| 3192 | + } | |
| 3193 | + } | |
| 3194 | + $keeper = $otto_tag ?: $all_tags[0]; | |
| 3195 | + | |
| 3196 | + # Callback form avoids backreference injection when the keyword list | |
| 3197 | + # contains $ followed by digits. | |
| 3198 | + $first_replaced = false; | |
| 3199 | + $html = preg_replace_callback($pattern, function ($m) use ($keeper, &$first_replaced) { | |
| 3200 | + if (!$first_replaced) { | |
| 3201 | + $first_replaced = true; | |
| 3202 | + return $keeper; | |
| 3203 | + } | |
| 3204 | + return ''; # Remove subsequent duplicates | |
| 3205 | + }, $html); | |
| 3206 | + | |
| 3207 | + return $html; | |
| 3208 | + } | |
| 3209 | + | |
| 3210 | + /** | |
| 2942 | 3211 | * Remove duplicate <link rel="canonical"> tags from HTML. |
| 2943 | 3212 | * |
| 2944 | 3213 | * When OTTO injects a canonical via header_html_insertion and MetaSync's |
| 2945 | 3214 | * SEO output (or WordPress core) has already emitted one, keep only the |
| @@ -2980,101 +3249,301 @@ | ||
| 2980 | 3249 | return $html; |
| 2981 | 3250 | } |
| 2982 | 3251 | |
| 2983 | 3252 | /** |
| 3253 | + * Normalize a JSON-LD `@type` into a list of lowercased type tokens. | |
| 3254 | + * | |
| 3255 | + * JSON-LD allows `@type` to be a string OR an array of strings, and Rank Math | |
| 3256 | + * routinely emits multi-typed nodes (e.g. ["WebPage", "CollectionPage"]). | |
| 3257 | + * Collapsing such a node to its first token hides it from a plain-typed | |
| 3258 | + * duplicate, so every token is returned and indexed separately. | |
| 3259 | + * | |
| 3260 | + * @param mixed $raw The raw `@type` value. | |
| 3261 | + * @return string[] Unique lowercased tokens, possibly empty. | |
| 3262 | + */ | |
| 3263 | + private function schema_type_tokens($raw) { | |
| 3264 | + $tokens = []; | |
| 3265 | + | |
| 3266 | + foreach ((is_array($raw) ? $raw : [$raw]) as $type) { | |
| 3267 | + if (is_array($type) || is_object($type) || is_bool($type) || $type === null) { | |
| 3268 | + continue; | |
| 3269 | + } | |
| 3270 | + $type = strtolower(trim((string) $type)); | |
| 3271 | + if ($type === '') { | |
| 3272 | + continue; | |
| 3273 | + } | |
| 3274 | + $tokens[$type] = true; | |
| 3275 | + } | |
| 3276 | + | |
| 3277 | + return array_keys($tokens); | |
| 3278 | + } | |
| 3279 | + | |
| 3280 | + /** | |
| 2984 | 3281 | * Deduplicate JSON-LD schema blocks. |
| 2985 | 3282 | * |
| 2986 | - * When OTTO and a third-party SEO plugin both inject <script type="application/ld+json"> | |
| 2987 | - * blocks, keep OTTO's version for any @type that appears in both. | |
| 2988 | - * Third-party blocks whose @type is not covered by OTTO are preserved. | |
| 3283 | + * Two nodes collide when they share an `@id`, or when one claims an `@type` | |
| 3284 | + * token the page already carries. Collisions resolve strictly in OTTO's | |
| 3285 | + * favour: | |
| 2989 | 3286 | * |
| 3287 | + * - an OTTO node (marked `data-otto`) displaces every colliding | |
| 3288 | + * third-party node, whichever appeared first in the document; | |
| 3289 | + * - two OTTO nodes describing the same entity keep the first; | |
| 3290 | + * - two third-party nodes never evict each other, even when they share a | |
| 3291 | + * type: a Yoast graph legitimately carries two distinct ImageObjects, | |
| 3292 | + * and MetaSync's own breadcrumb / local-business blocks are third-party | |
| 3293 | + * output this pass has no mandate to fold. | |
| 3294 | + * | |
| 3295 | + * When OTTO contributed no node at all the document comes back untouched — | |
| 3296 | + * a third-party-only page is not ours to rewrite. Edits are offset splices | |
| 3297 | + * into the original document, so a block that loses nothing keeps its exact | |
| 3298 | + * bytes (attributes, `class`, `@context`, position — head, body or footer) | |
| 3299 | + * and nothing is ever relocated. | |
| 3300 | + * | |
| 3301 | + * Blocks are left exactly where they are when their JSON does not parse, | |
| 3302 | + * when they carry neither `@type` nor `@graph`, or when they sit inside an | |
| 3303 | + * HTML comment or <noscript> — a schema node the site disabled by wrapping | |
| 3304 | + * it in a comment must stay disabled, not be re-published by this pass. | |
| 3305 | + * | |
| 2990 | 3306 | * @param string $html Full HTML. |
| 2991 | 3307 | * @return string |
| 2992 | 3308 | */ |
| 2993 | 3309 | private function deduplicate_schema_tags($html) { |
| 2994 | - // Find all JSON-LD script blocks | |
| 2995 | - $pattern = '/<script(\s[^>]*)type\s*=\s*(["\'])application\/ld\+json\2[^>]*>\s*([\s\S]*?)<\/script>/i'; | |
| 2996 | - if (preg_match_all($pattern, $html, $matches, PREG_SET_ORDER) <= 1) { | |
| 3310 | + // Find all JSON-LD script blocks. Group 1 is the FULL attribute string — | |
| 3311 | + // capturing only the part before `type=` would hide a `data-otto` marker | |
| 3312 | + // that sits after it, which is where every insertion path puts it. The | |
| 3313 | + // type VALUE is prefix-matched so `application/ld+json; charset=UTF-8` | |
| 3314 | + // is recognised too. | |
| 3315 | + $pattern = '/<script(\s[^>]*type\s*=\s*(["\'])application\/ld\+json[^"\'>]*\2[^>]*)>\s*([\s\S]*?)<\/script>/i'; | |
| 3316 | + if (!preg_match_all($pattern, $html, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) { | |
| 2997 | 3317 | return $html; |
| 2998 | 3318 | } |
| 2999 | 3319 | |
| 3000 | - $otto_by_type = []; // @type => decoded JSON object | |
| 3001 | - $third_by_type = []; // @type => decoded JSON object | |
| 3002 | - $otto_graph = []; // entries from OTTO @graph blocks | |
| 3003 | - $third_graph = []; // entries from third-party @graph blocks | |
| 3320 | + // Byte ranges this pass must not reach into. A commented-out (or | |
| 3321 | + // <noscript>-wrapped) schema block is deliberately disabled output. | |
| 3322 | + $masked = []; | |
| 3323 | + if (preg_match_all('/<!--[\s\S]*?-->|<noscript\b[^>]*>[\s\S]*?<\/noscript>/i', $html, $comments, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) { | |
| 3324 | + foreach ($comments as $comment) { | |
| 3325 | + $masked[] = [$comment[0][1], $comment[0][1] + strlen($comment[0][0])]; | |
| 3326 | + } | |
| 3327 | + } | |
| 3004 | 3328 | |
| 3329 | + // `data-otto` must be a whole attribute: `data-otto="true"` or a bare | |
| 3330 | + // `data-otto`. A substring test would also catch a hypothetical | |
| 3331 | + // `data-otto-thing="x"` and misattribute that block to OTTO. | |
| 3332 | + $is_otto_attr = '/(^|\s)data-otto(\s*=|\s|$)/i'; | |
| 3333 | + | |
| 3334 | + // Blocks able to take part, and their entities in document order. An | |
| 3335 | + // entity is one addressable node: a flat block's document, one | |
| 3336 | + // `@graph` entry, or one node of a top-level array. | |
| 3337 | + $blocks = []; | |
| 3338 | + $entities = []; | |
| 3339 | + | |
| 3005 | 3340 | foreach ($matches as $m) { |
| 3006 | - $attrs = $m[1]; | |
| 3007 | - $json_str = $m[3]; | |
| 3341 | + $start = $m[0][1]; | |
| 3342 | + foreach ($masked as $range) { | |
| 3343 | + if ($start >= $range[0] && $start < $range[1]) { | |
| 3344 | + continue 2; | |
| 3345 | + } | |
| 3346 | + } | |
| 3347 | + | |
| 3348 | + $attrs = $m[1][0]; | |
| 3349 | + $json_str = $m[3][0]; | |
| 3008 | 3350 | $decoded = json_decode($json_str, true); |
| 3009 | 3351 | if (json_last_error() !== JSON_ERROR_NONE || !is_array($decoded)) { |
| 3010 | - continue; // skip unparseable blocks — leave them in place | |
| 3352 | + continue; // unparseable — leave the block in place | |
| 3011 | 3353 | } |
| 3012 | - $is_otto = stripos($attrs, 'data-otto') !== false; | |
| 3013 | 3354 | |
| 3355 | + $is_otto = preg_match($is_otto_attr, $attrs) === 1; | |
| 3356 | + $end = $start + strlen($m[0][0]); | |
| 3357 | + | |
| 3358 | + // A trailing newline swallowed on full removal keeps the output tidy. | |
| 3359 | + $end_ws = $end; | |
| 3360 | + if (substr($html, $end, 2) === "\r\n") { | |
| 3361 | + $end_ws = $end + 2; | |
| 3362 | + } elseif (substr($html, $end, 1) === "\n") { | |
| 3363 | + $end_ws = $end + 1; | |
| 3364 | + } | |
| 3365 | + | |
| 3366 | + $register = function (array $node) use (&$entities, $is_otto) { | |
| 3367 | + // Keyed like metasync_otto_structured_entity_key(): @id first, | |
| 3368 | + // @type tokens for the type fold. Entries with neither stay in | |
| 3369 | + // their block but cannot collide, so they keep entity === null. | |
| 3370 | + $types = isset($node['@type']) ? $this->schema_type_tokens($node['@type']) : []; | |
| 3371 | + $id = (!empty($node['@id']) && is_string($node['@id'])) ? '@id:' . strtolower(trim($node['@id'])) : ''; | |
| 3372 | + if ($types === [] && $id === '') { | |
| 3373 | + return null; | |
| 3374 | + } | |
| 3375 | + $entities[] = ['otto' => $is_otto, 'types' => $types, 'id' => $id, 'alive' => true]; | |
| 3376 | + return count($entities) - 1; | |
| 3377 | + }; | |
| 3378 | + | |
| 3379 | + $block = ['start' => $start, 'end' => $end, 'end_ws' => $end_ws, 'attrs' => $attrs, 'decoded' => $decoded, 'items' => []]; | |
| 3380 | + | |
| 3014 | 3381 | if (isset($decoded['@graph']) && is_array($decoded['@graph'])) { |
| 3015 | - foreach ($decoded['@graph'] as $entry) { | |
| 3016 | - if (!isset($entry['@type'])) continue; | |
| 3017 | - // JSON-LD allows @type to be a string OR an array of strings. | |
| 3018 | - // Rank Math routinely emits multi-typed entries (e.g. ["Person", "Organization"]). | |
| 3019 | - // Using an array as an offset throws a fatal on PHP 8+, so normalize to scalar. | |
| 3020 | - $type = is_array($entry['@type']) | |
| 3021 | - ? (string) reset($entry['@type']) | |
| 3022 | - : (string) $entry['@type']; | |
| 3023 | - if ($type === '') continue; | |
| 3024 | - if ($is_otto) { | |
| 3025 | - $otto_graph[$type] = $entry; | |
| 3026 | - } else { | |
| 3027 | - $third_graph[$type] = $entry; | |
| 3028 | - } | |
| 3382 | + foreach ($decoded['@graph'] as $node) { | |
| 3383 | + $block['items'][] = ['node' => $node, 'e' => is_array($node) ? $register($node) : null]; | |
| 3029 | 3384 | } |
| 3385 | + $block['shape'] = 'graph'; | |
| 3030 | 3386 | } elseif (isset($decoded['@type'])) { |
| 3031 | - $type = is_array($decoded['@type']) | |
| 3032 | - ? (string) reset($decoded['@type']) | |
| 3033 | - : (string) $decoded['@type']; | |
| 3034 | - if ($type === '') continue; | |
| 3035 | - if ($is_otto) { | |
| 3036 | - $otto_by_type[$type] = $decoded; | |
| 3387 | + $block['shape'] = 'flat'; | |
| 3388 | + $block['items'][] = ['node' => $decoded, 'e' => $register($decoded)]; | |
| 3389 | + } elseif (isset($decoded[0])) { | |
| 3390 | + // A bare top-level array: [{"@type":"WebPage",...}, ...]. The | |
| 3391 | + // persistence path already folds this shape; the render path | |
| 3392 | + // used to skip it entirely, leaving its duplicates standing. | |
| 3393 | + $block['shape'] = 'list'; | |
| 3394 | + foreach ($decoded as $node) { | |
| 3395 | + $block['items'][] = ['node' => $node, 'e' => is_array($node) ? $register($node) : null]; | |
| 3396 | + } | |
| 3397 | + } else { | |
| 3398 | + continue; // neither @type nor @graph nor array — nothing to key on | |
| 3399 | + } | |
| 3400 | + | |
| 3401 | + $blocks[] = $block; | |
| 3402 | + } | |
| 3403 | + | |
| 3404 | + foreach ($entities as $entity) { | |
| 3405 | + if ($entity['otto']) { | |
| 3406 | + $otto_present = true; | |
| 3407 | + break; | |
| 3408 | + } | |
| 3409 | + } | |
| 3410 | + if (empty($otto_present)) { | |
| 3411 | + return $html; // OTTO contributed no schema — a third-party page is not ours to rewrite | |
| 3412 | + } | |
| 3413 | + | |
| 3414 | + // Fold. Type claims map a token to EVERY alive holder, so a later OTTO | |
| 3415 | + // node can displace a whole set of third-party duplicates at once. | |
| 3416 | + /** @var array<string, int> $by_id */ | |
| 3417 | + $by_id = []; // '@id:...' => entity index | |
| 3418 | + /** @var array<string, int[]> $by_type */ | |
| 3419 | + $by_type = []; // type token => entity index[] | |
| 3420 | + | |
| 3421 | + $release = function ($i) use (&$by_id, &$by_type, &$entities) { | |
| 3422 | + foreach ($entities[$i]['types'] as $token) { | |
| 3423 | + $holders = $by_type[$token] ?? []; | |
| 3424 | + $holders = array_values(array_diff($holders, [$i])); | |
| 3425 | + if ($holders === []) { | |
| 3426 | + unset($by_type[$token]); | |
| 3037 | 3427 | } else { |
| 3038 | - $third_by_type[$type] = $decoded; | |
| 3428 | + $by_type[$token] = $holders; | |
| 3039 | 3429 | } |
| 3040 | 3430 | } |
| 3431 | + $id = $entities[$i]['id']; | |
| 3432 | + if ($id !== '' && ($by_id[$id] ?? -1) === $i) { | |
| 3433 | + unset($by_id[$id]); | |
| 3434 | + } | |
| 3435 | + }; | |
| 3436 | + | |
| 3437 | + foreach ($entities as $i => $entity) { | |
| 3438 | + $alive = true; | |
| 3439 | + | |
| 3440 | + // The same real-world entity twice, whatever @type each copy carries. | |
| 3441 | + if ($entity['id'] !== '' && isset($by_id[$entity['id']])) { | |
| 3442 | + $occupant = $by_id[$entity['id']]; | |
| 3443 | + if ($entity['otto'] && !$entities[$occupant]['otto']) { | |
| 3444 | + $entities[$occupant]['alive'] = false; | |
| 3445 | + $release($occupant); | |
| 3446 | + unset($by_id[$entity['id']]); // re-claimed below if the entity survives | |
| 3447 | + } else { | |
| 3448 | + $alive = false; // first copy stands (OTTO's, or the earlier OTTO's) | |
| 3449 | + } | |
| 3450 | + } | |
| 3451 | + | |
| 3452 | + if ($alive) { | |
| 3453 | + foreach ($entity['types'] as $token) { | |
| 3454 | + $holders = $by_type[$token] ?? []; | |
| 3455 | + if ($holders === []) { | |
| 3456 | + continue; | |
| 3457 | + } | |
| 3458 | + $otto_holder = false; | |
| 3459 | + foreach ($holders as $holder) { | |
| 3460 | + if ($entities[$holder]['otto']) { | |
| 3461 | + $otto_holder = true; | |
| 3462 | + break; | |
| 3463 | + } | |
| 3464 | + } | |
| 3465 | + if ($otto_holder) { | |
| 3466 | + $alive = false; // an OTTO node already holds this token | |
| 3467 | + break; | |
| 3468 | + } | |
| 3469 | + if ($entity['otto']) { | |
| 3470 | + // OTTO displaces every third-party holder of the token and | |
| 3471 | + // releases ALL their claims, not just this one — otherwise | |
| 3472 | + // tokens the evicted node held alone would stay blocked and | |
| 3473 | + // later, legitimate nodes of that type would lose. | |
| 3474 | + foreach ($holders as $holder) { | |
| 3475 | + $entities[$holder]['alive'] = false; | |
| 3476 | + $release($holder); | |
| 3477 | + } | |
| 3478 | + } | |
| 3479 | + // Third party meeting third-party holders: both stand. | |
| 3480 | + } | |
| 3481 | + } | |
| 3482 | + | |
| 3483 | + if ($alive) { | |
| 3484 | + foreach ($entity['types'] as $token) { | |
| 3485 | + $by_type[$token][] = $i; | |
| 3486 | + } | |
| 3487 | + if ($entity['id'] !== '') { | |
| 3488 | + $by_id[$entity['id']] = $i; | |
| 3489 | + } | |
| 3490 | + } else { | |
| 3491 | + $entities[$i]['alive'] = false; | |
| 3492 | + } | |
| 3041 | 3493 | } |
| 3042 | 3494 | |
| 3043 | - // If OTTO provided no schema at all, nothing to deduplicate | |
| 3044 | - if (empty($otto_by_type) && empty($otto_graph)) { | |
| 3045 | - return $html; | |
| 3495 | + $any_evicted = false; | |
| 3496 | + foreach ($entities as $entity) { | |
| 3497 | + if (!$entity['alive']) { | |
| 3498 | + $any_evicted = true; | |
| 3499 | + break; | |
| 3500 | + } | |
| 3046 | 3501 | } |
| 3502 | + if (!$any_evicted) { | |
| 3503 | + return $html; // no collision — every original byte stays | |
| 3504 | + } | |
| 3047 | 3505 | |
| 3048 | - // Remove all JSON-LD blocks from HTML | |
| 3049 | - $html = preg_replace($pattern, '', $html); | |
| 3506 | + // Rebuild only the blocks that lost something, back-to-front so the | |
| 3507 | + // recorded offsets of earlier blocks stay valid. | |
| 3508 | + $edits = []; | |
| 3509 | + foreach ($blocks as $block) { | |
| 3510 | + $kept = []; | |
| 3511 | + $lost = 0; | |
| 3512 | + foreach ($block['items'] as $item) { | |
| 3513 | + if ($item['e'] === null || $entities[$item['e']]['alive']) { | |
| 3514 | + $kept[] = $item['node']; | |
| 3515 | + } else { | |
| 3516 | + $lost++; | |
| 3517 | + } | |
| 3518 | + } | |
| 3519 | + if ($lost === 0) { | |
| 3520 | + continue; | |
| 3521 | + } | |
| 3522 | + if ($kept === []) { | |
| 3523 | + $edits[] = [$block['start'], $block['end_ws'], '']; | |
| 3524 | + continue; | |
| 3525 | + } | |
| 3050 | 3526 | |
| 3051 | - // Re-insert flat (non-@graph) blocks: OTTO wins for matching @type | |
| 3052 | - $kept = array_merge($otto_by_type, array_diff_key($third_by_type, $otto_by_type)); | |
| 3053 | - $rebuilt = ''; | |
| 3054 | - foreach ($kept as $decoded) { | |
| 3055 | - $json = metasync_safe_json_ld_encode($decoded, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE); | |
| 3527 | + if ($block['shape'] === 'list') { | |
| 3528 | + $payload = $kept; | |
| 3529 | + } else { // graph — a flat block that lost its only entity is removed above | |
| 3530 | + $payload = $block['decoded']; | |
| 3531 | + $payload['@graph'] = $kept; | |
| 3532 | + } | |
| 3533 | + | |
| 3534 | + $json = metasync_safe_json_ld_encode($payload, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE); | |
| 3056 | 3535 | if (!is_string($json) || $json === '') { |
| 3057 | - continue; // never re-emit a block whose payload failed to encode | |
| 3536 | + continue; // never drop a block whose payload failed to re-encode | |
| 3058 | 3537 | } |
| 3059 | - $rebuilt .= '<script type="application/ld+json" data-otto="true">' . $json . "</script>\n"; | |
| 3060 | - } | |
| 3061 | 3538 | |
| 3062 | - // Re-insert merged @graph block (if any entries exist) | |
| 3063 | - $merged_graph = array_merge($third_graph, $otto_graph); // OTTO wins on duplicate @type | |
| 3064 | - if (!empty($merged_graph)) { | |
| 3065 | - $graph_obj = ['@context' => 'https://schema.org', '@graph' => array_values($merged_graph)]; | |
| 3066 | - $graph_json = metasync_safe_json_ld_encode($graph_obj, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE); | |
| 3067 | - if (is_string($graph_json) && $graph_json !== '') { | |
| 3068 | - $rebuilt .= '<script type="application/ld+json" data-otto="true">' . $graph_json . "</script>\n"; | |
| 3069 | - } | |
| 3539 | + // The original attribute string is reused verbatim so class, id and | |
| 3540 | + // the data-otto marker itself survive the rebuild. | |
| 3541 | + $edits[] = [$block['start'], $block['end'], '<script' . $block['attrs'] . '>' . $json . '</script>']; | |
| 3070 | 3542 | } |
| 3071 | 3543 | |
| 3072 | - // Re-inject before </head> | |
| 3073 | - if (!empty($rebuilt)) { | |
| 3074 | - $html = preg_replace_callback('/(<\/head>)/i', function ($m) use ($rebuilt) { | |
| 3075 | - return $rebuilt . $m[1]; | |
| 3076 | - }, $html, 1); | |
| 3544 | + foreach (array_reverse($edits) as $edit) { | |
| 3545 | + $html = substr($html, 0, $edit[0]) . $edit[2] . substr($html, $edit[1]); | |
| 3077 | 3546 | } |
| 3078 | 3547 | |
| 3079 | 3548 | return $html; |
| 3080 | 3549 | } |
| @@ -3093,14 +3562,14 @@ | ||
| 3093 | 3562 | return true; |
| 3094 | 3563 | } |
| 3095 | 3564 | |
| 3096 | 3565 | # Check if amp=1 query parameter is present |
| 3097 | - if (isset($_GET['amp']) && $_GET['amp'] == '1') { | |
| 3566 | + if (isset($_GET['amp']) && $_GET['amp'] == '1') { // phpcs:ignore WordPress.Security.NonceVerification.Recommended -- read-only AMP detection | |
| 3098 | 3567 | return true; |
| 3099 | 3568 | } |
| 3100 | 3569 | |
| 3101 | 3570 | # Check for other common AMP query parameters |
| 3102 | - if (isset($_GET['amp']) && !empty($_GET['amp'])) { | |
| 3571 | + if (isset($_GET['amp']) && !empty($_GET['amp'])) { // phpcs:ignore WordPress.Security.NonceVerification.Recommended -- read-only AMP detection | |
| 3103 | 3572 | return true; |
| 3104 | 3573 | } |
| 3105 | 3574 | |
| 3106 | 3575 | return false; |
| @@ -3105,8 +3574,32 @@ | ||
| 3105 | 3574 | |
| 3106 | 3575 | return false; |
| 3107 | 3576 | } |
| 3108 | 3577 | |
| 3578 | + # function to mark OTTO's own JSON-LD blocks so deduplicate_schema_tags() can | |
| 3579 | + # tell them apart from a third-party SEO plugin's blocks. Every insertion path | |
| 3580 | + # (DOM, string fallback, HTTP fallback) runs the fragment through this so the | |
| 3581 | + # bytes they insert are identical and the "already applied" checks still match. | |
| 3582 | + private function stamp_otto_json_ld($html){ | |
| 3583 | + | |
| 3584 | + $stamped = preg_replace_callback( | |
| 3585 | + '/<script(\s[^>]*type\s*=\s*(["\'])application\/ld\+json[^"\'>]*\2[^>]*)>/i', | |
| 3586 | + function ($m) { | |
| 3587 | + # idempotent: a block already carrying a data-otto attribute — | |
| 3588 | + # anywhere in the tag, not just before `type` — is left alone, | |
| 3589 | + # never given a second one | |
| 3590 | + if (preg_match('/(^|\s)data-otto(\s|=|$)/i', $m[1])) { | |
| 3591 | + return $m[0]; | |
| 3592 | + } | |
| 3593 | + return '<script' . $m[1] . ' data-otto="true">'; | |
| 3594 | + }, | |
| 3595 | + $html | |
| 3596 | + ); | |
| 3597 | + | |
| 3598 | + # preg_replace returns null on a PCRE failure — never hand back less than we got | |
| 3599 | + return is_string($stamped) ? $stamped : $html; | |
| 3600 | + } | |
| 3601 | + | |
| 3109 | 3602 | # this function insterts header html to the dom |
| 3110 | 3603 | function insert_header_html($data){ |
| 3111 | 3604 | |
| 3112 | 3605 | # check that we have the header html |
| @@ -3130,9 +3623,13 @@ | ||
| 3130 | 3623 | $marker = $is_amp_page ? '' : self::HEAD_OPTIMIZED_MARKER; |
| 3131 | 3624 | |
| 3132 | 3625 | # Escape JSON-LD payload closers so a `</script>` inside a JSON |
| 3133 | 3626 | # string value cannot terminate the element early (XSS/page corruption). |
| 3134 | - $escaped_header_insertion = metasync_escape_json_ld_blocks_in_html($data['header_html_insertion']); | |
| 3627 | + # Then stamp the blocks as OTTO's, exactly as the string and HTTP | |
| 3628 | + # fallbacks do, so the deduper can attribute them. | |
| 3629 | + $escaped_header_insertion = $this->stamp_otto_json_ld( | |
| 3630 | + metasync_escape_json_ld_blocks_in_html($data['header_html_insertion']) | |
| 3631 | + ); | |
| 3135 | 3632 | |
| 3136 | 3633 | $head->outertext = '<head>' . $marker . $escaped_header_insertion . $head->innertext . '</head>'; |
| 3137 | 3634 | |
| 3138 | 3635 | } |
| @@ -3516,18 +4013,16 @@ | ||
| 3516 | 4013 | } |
| 3517 | 4014 | |
| 3518 | 4015 | # Apply header HTML insertion (for schema, etc.) — only if DOM insertion didn't already apply it |
| 3519 | 4016 | if (!empty($replacement_data['header_html_insertion'])) { |
| 3520 | - # Escape JSON-LD payload closers first; the DOM path inserts the | |
| 3521 | - # same escaped bytes, so the already-applied check must match them. | |
| 3522 | - $escaped_header_insertion = metasync_escape_json_ld_blocks_in_html($replacement_data['header_html_insertion']); | |
| 3523 | - $header_html_check = trim($escaped_header_insertion); | |
| 4017 | + # Escape JSON-LD payload closers and stamp OTTO's blocks first; the DOM | |
| 4018 | + # path inserts the same bytes, so the already-applied check must compare | |
| 4019 | + # against the stamped form or it would insert a second copy. | |
| 4020 | + $header_html_insertion = $this->stamp_otto_json_ld( | |
| 4021 | + metasync_escape_json_ld_blocks_in_html($replacement_data['header_html_insertion']) | |
| 4022 | + ); | |
| 4023 | + $header_html_check = trim($header_html_insertion); | |
| 3524 | 4024 | if (strpos($result_html, $header_html_check) === false) { |
| 3525 | - $header_html_insertion = preg_replace( | |
| 3526 | - '/<script(\s[^>]*)type\s*=\s*(["\'])application\/ld\+json\2/i', | |
| 3527 | - '<script$1type=$2application/ld+json$2 data-otto="true"', | |
| 3528 | - $escaped_header_insertion | |
| 3529 | - ); | |
| 3530 | 4025 | $header_html = str_replace(array('\\', '$'), array('\\\\', '\\$'), $header_html_insertion); |
| 3531 | 4026 | # Insert before </head> |
| 3532 | 4027 | $result_html = preg_replace('/(<\/head>)/i', $header_html . "\n" . '$1', $result_html, 1); |
| 3533 | 4028 | } |
| @@ -3599,9 +4094,9 @@ | ||
| 3599 | 4094 | |
| 3600 | 4095 | $result_html = preg_replace_callback( |
| 3601 | 4096 | '/(<' . $heading_type . '(?:\s[^>]*)?>)(.*?)(<\/' . $heading_type . '>)/is', |
| 3602 | 4097 | function ($m) use ($current_value, $recommended_value) { |
| 3603 | - $inner_text = trim(preg_replace('/\s+/', ' ', html_entity_decode(strip_tags($m[2]), ENT_QUOTES, 'UTF-8'))); | |
| 4098 | + $inner_text = trim(preg_replace('/\s+/', ' ', html_entity_decode(wp_strip_all_tags($m[2]), ENT_QUOTES, 'UTF-8'))); | |
| 3604 | 4099 | if ($inner_text !== $current_value) { |
| 3605 | 4100 | return $m[0]; |
| 3606 | 4101 | } |
| 3607 | 4102 | # Keep the author's inline formatting; skip when it can't |
| @@ -3659,8 +4154,9 @@ | ||
| 3659 | 4154 | |
| 3660 | 4155 | # DEDUPLICATION: Remove duplicate <title>, meta description, OG, Twitter tags, canonical, and JSON-LD schema |
| 3661 | 4156 | $result_html = $this->otto_guard_html($this->deduplicate_title_tags($result_html), $result_html, 'deduplicate_title_tags'); |
| 3662 | 4157 | $result_html = $this->otto_guard_html($this->deduplicate_description_tags($result_html), $result_html, 'deduplicate_description_tags'); |
| 4158 | + $result_html = $this->otto_guard_html($this->deduplicate_keywords_tags($result_html), $result_html, 'deduplicate_keywords_tags'); | |
| 3663 | 4159 | $result_html = $this->otto_guard_html($this->deduplicate_og_twitter_tags($result_html), $result_html, 'deduplicate_og_twitter_tags'); |
| 3664 | 4160 | $result_html = $this->otto_guard_html($this->apply_metabox_og_precedence($result_html), $result_html, 'apply_metabox_og_precedence'); |
| 3665 | 4161 | $result_html = $this->otto_guard_html($this->apply_custom_seo_precedence($result_html), $result_html, 'apply_custom_seo_precedence'); |
| 3666 | 4162 | $result_html = $this->otto_guard_html($this->deduplicate_schema_tags($result_html), $result_html, 'deduplicate_schema_tags'); |
| @@ -3741,12 +4237,26 @@ | ||
| 3741 | 4237 | |
| 3742 | 4238 | return $result_html; |
| 3743 | 4239 | |
| 3744 | 4240 | } catch (Exception $e) { |
| 3745 | - error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': Exception in process_html_directly - ' . $e->getMessage()); | |
| 4241 | + if (function_exists('metasync_otto_report_render_failure')) { | |
| 4242 | + metasync_otto_report_render_failure( | |
| 4243 | + 'MODIFICATION_EXCEPTION', | |
| 4244 | + 'OTTO rewrite threw: ' . $e->getMessage(), | |
| 4245 | + ['exception' => get_class($e)], | |
| 4246 | + 'error' | |
| 4247 | + ); | |
| 4248 | + } | |
| 3746 | 4249 | return false; |
| 3747 | 4250 | } catch (Error $e) { |
| 3748 | - error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': Error in process_html_directly - ' . $e->getMessage()); | |
| 4251 | + if (function_exists('metasync_otto_report_render_failure')) { | |
| 4252 | + metasync_otto_report_render_failure( | |
| 4253 | + 'MODIFICATION_ERROR', | |
| 4254 | + 'OTTO rewrite errored: ' . $e->getMessage(), | |
| 4255 | + ['error' => get_class($e)], | |
| 4256 | + 'error' | |
| 4257 | + ); | |
| 4258 | + } | |
| 3749 | 4259 | return false; |
| 3750 | 4260 | } |
| 3751 | 4261 | } |
| 3752 | 4262 | |
| @@ -3795,23 +4305,15 @@ | ||
| 3795 | 4305 | } |
| 3796 | 4306 | } |
| 3797 | 4307 | } |
| 3798 | 4308 | |
| 3799 | - if ($remove_title && !$has_custom_title) { | |
| 3800 | - # Remove Open Graph and Twitter title tags (keep main <title>) | |
| 3801 | - $title_selectors = [ | |
| 3802 | - 'meta[property=og:title]', | |
| 3803 | - 'meta[name=twitter:title]', | |
| 3804 | - ]; | |
| 4309 | + # Preserve third-party og:title/twitter:title tags here. They used to be | |
| 4310 | + # stripped whenever $remove_title was set, but that also removes them | |
| 4311 | + # when no custom title exists to replace them, leaving pages with no | |
| 4312 | + # og:title/twitter:title at all. OTTO's own title/OG dedup and | |
| 4313 | + # precedence passes already reconcile duplicates, so this branch no | |
| 4314 | + # longer needs to delete those tags. | |
| 3805 | 4315 | |
| 3806 | - foreach ($title_selectors as $selector) { | |
| 3807 | - $tags = $this->dom->find($selector); | |
| 3808 | - foreach ($tags as $tag) { | |
| 3809 | - $tag->outertext = ''; # Remove the tag | |
| 3810 | - } | |
| 3811 | - } | |
| 3812 | - } | |
| 3813 | - | |
| 3814 | 4316 | # Save changes |
| 3815 | 4317 | $this->save_reload(); |
| 3816 | 4318 | } |
| 3817 | 4319 | |
| @@ -4035,5 +4537,5 @@ | ||
| 4035 | 4537 | return $html; |
| 4036 | 4538 | } |
| 4037 | 4539 | |
| 4038 | 4540 | |
| 4039 | -} | |
| 4541 | +} | |
