← All changes | includes/seo/class-analytics-manager.php +194 -35 2.12.0 → 2.14.2 View file →

		@@ -124,8 +124,45 @@
124	124	*/
125	125	private const REFRESH_TIMEOUT = 10;
126	126
127	127	/**
	128	+ * Transient prefix for the cached dashboard payload, suffixed with the
	129	+ * date range ("30d"). Carries the plugin prefix so uninstall's
	130	+ * `_transient_thinkrank_%` sweep removes it; the payload holds up to
	131	+ * 1,000 of the site's search queries (#918).
	132	+ *
	133	+ * @since 2.14.2
	134	+ * @var string
	135	+ */
	136	+ public const DASHBOARD_CACHE_PREFIX = 'thinkrank_analytics_dashboard_v5_';
	137	+
	138	+ /**
	139	+ * Transient prefix for the cached SEO opportunities payload, suffixed
	140	+ * with the date range. Prefixed for the same reason as the dashboard's.
	141	+ *
	142	+ * @since 2.14.2
	143	+ * @var string
	144	+ */
	145	+ public const OPPORTUNITIES_CACHE_PREFIX = 'thinkrank_seo_opportunities_';
	146	+
	147	+ /**
	148	+ * Every date range the dashboard cache is written under: the three the
	149	+ * UI offers plus the doubled previous-period ranges used for trends.
	150	+ *
	151	+ * @since 2.14.2
	152	+ * @var string[]
	153	+ */
	154	+ public const CACHED_DASHBOARD_RANGES = ['7d', '30d', '90d', '14d', '60d', '180d'];
	155	+
	156	+ /**
	157	+ * Every date range the opportunities cache is written under.
	158	+ *
	159	+ * @since 2.14.2
	160	+ * @var string[]
	161	+ */
	162	+ public const CACHED_OPPORTUNITY_RANGES = ['7d', '30d', '90d'];
	163	+
	164	+ /**
128	165	* Constructor
129	166	*
130	167	* @param Settings_Manager|null $settings_manager Settings manager instance
131	168	*/
		@@ -677,9 +714,9 @@
677	714	*
678	715	* @throws \Exception On failure.
679	716	*/
680	717	public function get_dashboard_data(string $date_range = '30d'): array {
681		- $cache_key = "analytics_dashboard_v5_{$date_range}";
	718	+ $cache_key = self::DASHBOARD_CACHE_PREFIX . $date_range;
682	719	$cached_data = get_transient($cache_key);
683	720
684	721	if ($cached_data !== false) {
685	722	// Core Web Vitals are cached separately with a much shorter
		@@ -750,29 +787,17 @@
750	787	// Fallback to old client if new one fails init (shouldn't happen if they use same creds)
751	788	$search_performance = $this->search_console_client->get_search_performance($site_url, $date_range, ['query'], 1000);
752	789	}
753	790
754		- // Calculate position distribution
755		- $position_distribution = [
756		- 'top_3' => 0,
757		- '4_10' => 0,
758		- '10_50' => 0,
759		- '51_100' => 0
760		- ];
	791	+ // Position distribution over every query with an
	792	+ // impression, not over the 1,000-row list above (#913).
	793	+ $position_distribution = $this->search_analytics_client
	794	+ ? $this->count_position_distribution($site_url, $start_date, $end_date, $search_performance['rows'] ?? [])
	795	+ : self::bucket_positions(
	796	+ $search_performance['rows'] ?? [],
	797	+ count($search_performance['rows'] ?? []) < 1000
	798	+ );
761	799
762		- foreach ($search_performance['rows'] ?? [] as $row) {
763		- $position = $row['position'] ?? 0;
764		- if ($position <= 3) {
765		- $position_distribution['top_3']++;
766		- } elseif ($position <= 10) {
767		- $position_distribution['4_10']++;
768		- } elseif ($position <= 50) {
769		- $position_distribution['10_50']++;
770		- } elseif ($position <= 100) {
771		- $position_distribution['51_100']++;
772		- }
773		- }
774		-
775	800	$dashboard_data['search_performance'] = array_merge($search_performance, [
776	801	'totals' => $totals,
777	802	'position_distribution' => $position_distribution
778	803	]);
		@@ -814,8 +839,132 @@
814	839	return $dashboard_data;
815	840	}
816	841
817	842	/**
	843	+ * Rows per page when counting the position distribution. The most the
	844	+ * Search Analytics API returns in one request.
	845	+ *
	846	+ * @since 2.15.0
	847	+ */
	848	+ private const POSITION_PAGE_SIZE = 25000;
	849	+
	850	+ /**
	851	+ * Pages read before the count stops: 200,000 queries. A property past
	852	+ * that is counted over its top 200,000 by clicks and flagged incomplete.
	853	+ *
	854	+ * @since 2.15.0
	855	+ */
	856	+ private const POSITION_MAX_PAGES = 8;
	857	+
	858	+ /**
	859	+ * Count the period's queries into position buckets across the whole
	860	+ * property (#913).
	861	+ *
	862	+ * The dashboard's query list is capped at 1,000 rows ordered by clicks,
	863	+ * so counting it told any larger site it had exactly 1,000 queries and
	864	+ * dropped the long tail, which is where positions 51-100 live. When that
	865	+ * list came back short it already holds every query and is counted as
	866	+ * is, with no extra request. Otherwise the property is paged with
	867	+ * `startRow` at POSITION_PAGE_SIZE rows until a short page, counting as
	868	+ * rows arrive rather than keeping them.
	869	+ *
	870	+ * A page that fails (other than a 401, which is re-thrown so the token
	871	+ * refresh runs) leaves the count at what was read so far, flagged
	872	+ * `complete: false`, rather than failing the whole dashboard.
	873	+ *
	874	+ * @since 2.15.0
	875	+ *
	876	+ * @param string $site_url Search Console property.
	877	+ * @param string $start_date Window start (Y-m-d).
	878	+ * @param string $end_date Window end (Y-m-d).
	879	+ * @param array $first_rows The capped query list already fetched.
	880	+ * @return array{top_3:int,4_10:int,10_50:int,51_100:int,over_100:int,complete:bool}
	881	+ * @throws \Exception On a 401, so get_dashboard_data() can refresh the token.
	882	+ */
	883	+ private function count_position_distribution(string $site_url, string $start_date, string $end_date, array $first_rows): array {
	884	+ if (count($first_rows) < 1000) {
	885	+ return self::bucket_positions($first_rows, true);
	886	+ }
	887	+
	888	+ $distribution = self::bucket_positions([], true);
	889	+ $start_row = 0;
	890	+
	891	+ for ($page = 0; $page < self::POSITION_MAX_PAGES; $page++) {
	892	+ try {
	893	+ $rows = $this->search_analytics_client->get_search_analytics_data(
	894	+ $site_url,
	895	+ $start_date,
	896	+ $end_date,
	897	+ ['query'],
	898	+ self::POSITION_PAGE_SIZE,
	899	+ $start_row
	900	+ )['rows'] ?? [];
	901	+ } catch (\Exception $e) {
	902	+ if ($e->getCode() === 401) {
	903	+ throw $e;
	904	+ }
	905	+ // Nothing read yet: the capped list is the best there is.
	906	+ $partial = $start_row === 0 ? self::bucket_positions($first_rows, false) : $distribution;
	907	+ $partial['complete'] = false;
	908	+ return $partial;
	909	+ }
	910	+
	911	+ $page_counts = self::bucket_positions($rows, true);
	912	+ foreach (['top_3', '4_10', '10_50', '51_100', 'over_100'] as $bucket) {
	913	+ $distribution[$bucket] += $page_counts[$bucket];
	914	+ }
	915	+
	916	+ if (count($rows) < self::POSITION_PAGE_SIZE) {
	917	+ return $distribution;
	918	+ }
	919	+ $start_row += self::POSITION_PAGE_SIZE;
	920	+ }
	921	+
	922	+ $distribution['complete'] = false;
	923	+ return $distribution;
	924	+ }
	925	+
	926	+ /**
	927	+ * Bucket Search Console rows by average position.
	928	+ *
	929	+ * `10_50` is the historical key for positions 11-50. Rows past 100 are
	930	+ * counted in `over_100`: they still had impressions.
	931	+ *
	932	+ * @since 2.15.0
	933	+ *
	934	+ * @param array $rows Search Console rows.
	935	+ * @param bool $complete Whether $rows is every query in the window.
	936	+ * @return array{top_3:int,4_10:int,10_50:int,51_100:int,over_100:int,complete:bool}
	937	+ */
	938	+ private static function bucket_positions(array $rows, bool $complete): array {
	939	+ $distribution = [
	940	+ 'top_3' => 0,
	941	+ '4_10' => 0,
	942	+ '10_50' => 0,
	943	+ '51_100' => 0,
	944	+ 'over_100' => 0,
	945	+ 'complete' => $complete,
	946	+ ];
	947	+
	948	+ foreach ($rows as $row) {
	949	+ $position = (float) ($row['position'] ?? 0);
	950	+ if ($position <= 3) {
	951	+ $distribution['top_3']++;
	952	+ } elseif ($position <= 10) {
	953	+ $distribution['4_10']++;
	954	+ } elseif ($position <= 50) {
	955	+ $distribution['10_50']++;
	956	+ } elseif ($position <= 100) {
	957	+ $distribution['51_100']++;
	958	+ } else {
	959	+ $distribution['over_100']++;
	960	+ }
	961	+ }
	962	+
	963	+ return $distribution;
	964	+ }
	965	+
	966	+ /**
818	967	* Get Core Web Vitals for the analytics dashboard, cached independently
819	968	* of the dashboard payload.
820	969	*
821	970	* Successful results are cached for 1 hour; failures are never cached
		@@ -858,9 +1007,9 @@
858	1007	* @param string $date_range Date range for analysis
859	1008	* @return array SEO opportunities
860	1009	*/
861	1010	public function get_seo_opportunities(string $date_range = '30d'): array {
862		- $cache_key = "seo_opportunities_{$date_range}";
	1011	+ $cache_key = self::OPPORTUNITIES_CACHE_PREFIX . $date_range;
863	1012	$cached_data = get_transient($cache_key);
864	1013
865	1014	if ($cached_data !== false) {
866	1015	return $cached_data;
		@@ -974,8 +1123,27 @@
974	1123	}
975	1124	}
976	1125
977	1126	/**
	1127	+ * Every transient the dashboard and opportunities caches are written
	1128	+ * under. One list, so "Refresh data", the settings save and any later
	1129	+ * caller clear the same keys the readers use.
	1130	+ *
	1131	+ * @since 2.14.2
	1132	+ * @return string[]
	1133	+ */
	1134	+ public static function dashboard_cache_keys(): array {
	1135	+ $keys = [];
	1136	+ foreach (self::CACHED_DASHBOARD_RANGES as $range) {
	1137	+ $keys[] = self::DASHBOARD_CACHE_PREFIX . $range;
	1138	+ }
	1139	+ foreach (self::CACHED_OPPORTUNITY_RANGES as $range) {
	1140	+ $keys[] = self::OPPORTUNITIES_CACHE_PREFIX . $range;
	1141	+ }
	1142	+ return $keys;
	1143	+ }
	1144	+
	1145	+ /**
978	1146	* Force refresh of all cached data
979	1147	*
980	1148	* @return array Refresh results
981	1149	*/
		@@ -982,21 +1150,12 @@
982	1150	public function refresh_data(): array {
983	1151	// Clear all analytics-related transients, including the previous-period
984	1152	// ranges used for trend comparison (14d/60d/180d) and the separately
985	1153	// cached Core Web Vitals payload.
986		- $cache_keys = [
987		- 'analytics_dashboard_v5_7d',
988		- 'analytics_dashboard_v5_30d',
989		- 'analytics_dashboard_v5_90d',
990		- 'analytics_dashboard_v5_14d',
991		- 'analytics_dashboard_v5_60d',
992		- 'analytics_dashboard_v5_180d',
993		- 'seo_opportunities_7d',
994		- 'seo_opportunities_30d',
995		- 'seo_opportunities_90d',
996		- 'indexing_status',
997		- 'thinkrank_dashboard_cwv'
998		- ];
	1154	+ $cache_keys = array_merge(
	1155	+ self::dashboard_cache_keys(),
	1156	+ ['indexing_status', 'thinkrank_dashboard_cwv']
	1157	+ );
999	1158
1000	1159	// Also clear PageSpeed-derived caches. Their keys are md5-derived from
1001	1160	// URL + device, so compute them for the URL/device combinations the
1002	1161	// plugin actually tests.