| @@ -7,24 +7,27 @@ | ||
| 7 | 7 | /** |
| 8 | 8 | * This class handles the Otto Pixel Functions |
| 9 | 9 | */ |
| 10 | 10 | Class Metasync_otto_pixel{ |
| 11 | - | |
| 11 | + | |
| 12 | 12 | #otto html class |
| 13 | 13 | public $o_html; |
| 14 | 14 | |
| 15 | - # no cache wp pages | |
| 15 | + # crawl data option | |
| 16 | + public $option_name = 'metasync_otto_crawldata'; | |
| 17 | + | |
| 18 | + # no cache wp pages | |
| 16 | 19 | public $no_cache_pages = ['wp-login.php']; |
| 17 | 20 | |
| 18 | 21 | # OTTO UUID |
| 19 | 22 | private $otto_uuid; |
| 20 | 23 | |
| 21 | - # | |
| 24 | + # | |
| 22 | 25 | function __construct($otto_uuid){ |
| 23 | - | |
| 26 | + | |
| 24 | 27 | # store the UUID |
| 25 | 28 | $this->otto_uuid = $otto_uuid; |
| 26 | - | |
| 29 | + | |
| 27 | 30 | # load the html class |
| 28 | 31 | $this->o_html = new Metasync_otto_html($otto_uuid); |
| 29 | 32 | |
| 30 | 33 | } |
| @@ -36,8 +39,102 @@ | ||
| 36 | 39 | # All pages are processed in real-time from OTTO API |
| 37 | 40 | return true; |
| 38 | 41 | } |
| 39 | 42 | |
| 43 | + | |
| 44 | + # method to save crawl data into | |
| 45 | + function save_crawl_data($data){ | |
| 46 | + | |
| 47 | + # get the option name | |
| 48 | + $option_name = $this->option_name; | |
| 49 | + | |
| 50 | + # handle data | |
| 51 | + $saved = get_option($option_name); | |
| 52 | + | |
| 53 | + # log saved | |
| 54 | + | |
| 55 | + # if saved false save | |
| 56 | + if(empty($saved['urls'])){ | |
| 57 | + | |
| 58 | + # save the option | |
| 59 | + update_option($option_name, $data); | |
| 60 | + | |
| 61 | + return; | |
| 62 | + } | |
| 63 | + | |
| 64 | + # get new unique list of urls | |
| 65 | + $new_list = array_unique(array_merge($data['urls'], $saved['urls'])); | |
| 66 | + | |
| 67 | + # set data | |
| 68 | + $data['urls'] = $new_list; | |
| 69 | + | |
| 70 | + # log saved | |
| 71 | + | |
| 72 | + # save the option | |
| 73 | + update_option($option_name, $data); | |
| 74 | + } | |
| 75 | + | |
| 76 | + /** | |
| 77 | + * Check if a URL has been crawled by Otto | |
| 78 | + * - Validates domain against saved domain | |
| 79 | + * - Ignores query strings | |
| 80 | + * - Does exact path matching as Otto stores paths exactly as they appear | |
| 81 | + * @param string $url - Full URL to check | |
| 82 | + * @return bool - True if URL was crawled, false otherwise | |
| 83 | + */ | |
| 84 | + function is_url_crawled($url) { | |
| 85 | + # Ensure valid input | |
| 86 | + if (empty($url) || !is_string($url)) { | |
| 87 | + return false; | |
| 88 | + } | |
| 89 | + | |
| 90 | + # Get saved crawl data | |
| 91 | + $saved = get_option($this->option_name); | |
| 92 | + | |
| 93 | + if ( | |
| 94 | + empty($saved) || | |
| 95 | + !is_array($saved) || | |
| 96 | + empty($saved['domain']) || | |
| 97 | + empty($saved['urls']) || | |
| 98 | + !is_array($saved['urls']) | |
| 99 | + ) { | |
| 100 | + return false; | |
| 101 | + } | |
| 102 | + | |
| 103 | + # Parse saved domain + incoming URL | |
| 104 | + $saved_domain = wp_parse_url($saved['domain'], PHP_URL_HOST); | |
| 105 | + $incoming_domain = wp_parse_url($url, PHP_URL_HOST); | |
| 106 | + | |
| 107 | + # Domain mismatch - not crawled | |
| 108 | + if (empty($saved_domain) || empty($incoming_domain) || strcasecmp($saved_domain, $incoming_domain) !== 0) { | |
| 109 | + return false; | |
| 110 | + } | |
| 111 | + | |
| 112 | + # Parse incoming path (ignore query string) | |
| 113 | + $parsed_url = wp_parse_url($url); | |
| 114 | + $url_path = $parsed_url['path'] ?? '/'; | |
| 115 | + | |
| 116 | + # Ensure path starts with / but don't modify trailing slashes | |
| 117 | + # Otto stores paths exactly as they appear in URLs | |
| 118 | + if (substr($url_path, 0, 1) !== '/') { | |
| 119 | + $url_path = '/' . $url_path; | |
| 120 | + } | |
| 121 | + | |
| 122 | + # Compare against crawled URLs - exact match | |
| 123 | + foreach ($saved['urls'] as $crawled_url) { | |
| 124 | + if (!is_string($crawled_url)) { | |
| 125 | + continue; | |
| 126 | + } | |
| 127 | + | |
| 128 | + # Direct comparison - Otto stores paths exactly as they are | |
| 129 | + if (strcasecmp($crawled_url, $url_path) === 0) { | |
| 130 | + return true; | |
| 131 | + } | |
| 132 | + } | |
| 133 | + | |
| 134 | + return false; | |
| 135 | + } | |
| 136 | + | |
| 40 | 137 | # get the current route |
| 41 | 138 | function get_route(){ |
| 42 | 139 | |
| 43 | 140 | # check if we're in an HTTP context |
| @@ -226,30 +323,8 @@ | ||
| 226 | 323 | |
| 227 | 324 | # Get cache status for headers |
| 228 | 325 | $cache_status = Metasync_Otto_Transient_Cache::get_cache_status($cache_track_key); |
| 229 | 326 | |
| 230 | - # POSTMETA BACKSTOP: a MISS means this request already paid for a live | |
| 231 | - # OTTO fetch, which makes it the cheapest possible place to repair the | |
| 232 | - # stored SEO meta when the crawl-notify webhook never ran. The sync is | |
| 233 | - # deferred to a shutdown callback that runs after the response HTML is | |
| 234 | - # built, so the rendered page and its timing are untouched. A HIT never | |
| 235 | - # reaches this branch — steady-state requests keep doing zero extra | |
| 236 | - # work. The class_exists()/method_exists() pair mirrors the guard | |
| 237 | - # above: during a plugin update an opcache can hold this file without | |
| 238 | - # the new class, and an unguarded static call would fatal the page. | |
| 239 | - # The whole hook sits in try/catch: in the mixed-opcache window a missing | |
| 240 | - # class constant is an Error, and the backstop must never fatal the page. | |
| 241 | - try { | |
| 242 | - $is_cache_miss = ($cache_status === Metasync_Otto_Transient_Cache::STATUS_MISS); | |
| 243 | - # @phpstan-ignore-next-line function.alreadyNarrowedType | |
| 244 | - if ($is_cache_miss && class_exists('Metasync_Otto_Backstop_Sync') && method_exists('Metasync_Otto_Backstop_Sync', 'maybe_defer_sync')) { | |
| 245 | - Metasync_Otto_Backstop_Sync::maybe_defer_sync($route, $suggestions, $cache_status); | |
| 246 | - } | |
| 247 | - } catch (Throwable $backstop_error) { | |
| 248 | - // Best-effort repair only; rendering continues untouched. | |
| 249 | - unset($backstop_error); | |
| 250 | - } | |
| 251 | - | |
| 252 | 327 | # RENDER PRIORITY: |
| 253 | 328 | # 1. WP Rocket rocket_buffer filter — WP Rocket caches OTTO-modified HTML, |
| 254 | 329 | # so Kinsta also caches the correct version. No exit(), no cache bypass. |
| 255 | 330 | # 2. Output Buffer (fast) — for environments without WP Rocket. |
| @@ -265,9 +340,9 @@ | ||
| 265 | 340 | # WP ROCKET PATH: highest priority when WP Rocket is active. |
| 266 | 341 | # rocket_buffer fires inside WP Rocket's own cache pipeline, so the |
| 267 | 342 | # cache file WP Rocket writes (and that Kinsta stores) is already |
| 268 | 343 | # OTTO-modified. Eliminates the race where WP Rocket saved pre-OTTO HTML. |
| 269 | - if (defined('WP_ROCKET_VERSION') || class_exists('WP_Rocket')) { | |
| 344 | + if (class_exists('WP_Rocket')) { | |
| 270 | 345 | $wp_rocket_compat_mode = Metasync_Otto_Config::get_wp_rocket_compat_mode(); |
| 271 | 346 | if ($wp_rocket_compat_mode !== 'http') { |
| 272 | 347 | # disable_otto exits in metasync_start_otto() before we get here, so |
| 273 | 348 | # only 'http' compat mode needs to skip the rocket_buffer path. |
| @@ -765,9 +840,8 @@ | ||
| 765 | 840 | Metasync_Otto_Render_Strategy::block_http_cache(); |
| 766 | 841 | } |
| 767 | 842 | |
| 768 | 843 | # continue to render the html |
| 769 | - // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- server-rendered route HTML from the OTTO render strategy; escaping would corrupt the document | |
| 770 | 844 | echo $route_html_string; |
| 771 | 845 | |
| 772 | 846 | # prevent further wp execution |
| 773 | 847 | exit(); |