← All changes | includes/seo/class-analytics-manager.php +249 -84 2.7.0 → 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	*/
		@@ -601,66 +638,72 @@
601	638	*
602	639	* @return array Connection test results
603	640	*/
604	641	public function test_connections(): array {
	642	+ // Every entry carries a `message`, configured or not: the result is
	643	+ // handed to the Integrations screen as JSON, and a key that exists on
	644	+ // some services and not others reads as `undefined` there.
605	645	$results = [
606		- 'google_analytics' => ['status' => 'not_configured'],
607		- 'search_console' => ['status' => 'not_configured'],
608		- 'pagespeed' => ['status' => 'not_configured']
	646	+ 'google_analytics' => ['status' => 'not_configured', 'message' => ''],
	647	+ 'search_console' => ['status' => 'not_configured', 'message' => ''],
	648	+ 'pagespeed' => ['status' => 'not_configured', 'message' => '']
609	649	];
610	650
611		- // Test Google Analytics connection
612		- if ($this->analytics_client) {
613		- try {
614		- $test_result = $this->analytics_client->test_connection();
615		- $results['google_analytics'] = [
616		- 'status' => $test_result['success'] ? 'connected' : 'error',
617		- 'message' => $test_result['message'],
618		- 'details' => $test_result
619		- ];
620		- } catch (\Exception $e) {
621		- $results['google_analytics'] = [
622		- 'status' => 'error',
623		- 'message' => $e->getMessage()
624		- ];
	651	+ // One shape for all three. They were three copies of the same block
	652	+ // differing only in the client, which is how the same unguarded read
	653	+ // came to exist in triplicate (#852).
	654	+ $clients = [
	655	+ 'google_analytics' => $this->analytics_client,
	656	+ 'search_console' => $this->search_console_client,
	657	+ 'pagespeed' => $this->pagespeed_client,
	658	+ ];
	659	+
	660	+ foreach ($clients as $service => $client) {
	661	+ if (!$client) {
	662	+ continue;
625	663	}
	664	+
	665	+ $results[$service] = $this->describe_connection_test($client);
626	666	}
627	667
628		- // Test Search Console connection
629		- if ($this->search_console_client) {
630		- try {
631		- $test_result = $this->search_console_client->test_connection();
632		- $results['search_console'] = [
633		- 'status' => $test_result['success'] ? 'connected' : 'error',
634		- 'message' => $test_result['message'],
635		- 'details' => $test_result
636		- ];
637		- } catch (\Exception $e) {
638		- $results['search_console'] = [
639		- 'status' => 'error',
640		- 'message' => $e->getMessage()
641		- ];
	668	+ return $results;
	669	+ }
	670	+
	671	+ /**
	672	+ * Run one client's connection test and report it in a fixed shape.
	673	+ *
	674	+ * The clients answer success with `message` and failure with `error`, and
	675	+ * this read `$test_result['message']` unconditionally — so a failed test
	676	+ * raised "Undefined array key" and handed the admin an error with
	677	+ * `message => null`. The reason the client had in hand was the one thing
	678	+ * the screen needed. The clients now set `message` on both branches; the
	679	+ * fallbacks here cover a client that does not, including one that answers
	680	+ * with neither key.
	681	+ *
	682	+ * @since 2.12.0
	683	+ *
	684	+ * @param object $client A client exposing test_connection(): array.
	685	+ * @return array{status:string, message:string, details?:array}
	686	+ */
	687	+ private function describe_connection_test($client): array {
	688	+ try {
	689	+ $test_result = $client->test_connection();
	690	+
	691	+ if (!is_array($test_result)) {
	692	+ return ['status' => 'error', 'message' => ''];
642	693	}
643		- }
644	694
645		- // Test PageSpeed connection
646		- if ($this->pagespeed_client) {
647		- try {
648		- $test_result = $this->pagespeed_client->test_connection();
649		- $results['pagespeed'] = [
650		- 'status' => $test_result['success'] ? 'connected' : 'error',
651		- 'message' => $test_result['message'],
652		- 'details' => $test_result
653		- ];
654		- } catch (\Exception $e) {
655		- $results['pagespeed'] = [
656		- 'status' => 'error',
657		- 'message' => $e->getMessage()
658		- ];
659		- }
	695	+ return [
	696	+ 'status' => !empty($test_result['success']) ? 'connected' : 'error',
	697	+ 'message' => (string) ($test_result['message'] ?? $test_result['error'] ?? ''),
	698	+ 'details' => $test_result,
	699	+ ];
	700	+ } catch (\Exception $e) {
	701	+ return [
	702	+ 'status' => 'error',
	703	+ 'message' => $e->getMessage(),
	704	+ ];
660	705	}
661		-
662		- return $results;
663	706	}
664	707
665	708	/**
666	709	* Get analytics dashboard data
		@@ -671,9 +714,9 @@
671	714	*
672	715	* @throws \Exception On failure.
673	716	*/
674	717	public function get_dashboard_data(string $date_range = '30d'): array {
675		- $cache_key = "analytics_dashboard_v5_{$date_range}";
	718	+ $cache_key = self::DASHBOARD_CACHE_PREFIX . $date_range;
676	719	$cached_data = get_transient($cache_key);
677	720
678	721	if ($cached_data !== false) {
679	722	// Core Web Vitals are cached separately with a much shorter
		@@ -744,29 +787,17 @@
744	787	// Fallback to old client if new one fails init (shouldn't happen if they use same creds)
745	788	$search_performance = $this->search_console_client->get_search_performance($site_url, $date_range, ['query'], 1000);
746	789	}
747	790
748		- // Calculate position distribution
749		- $position_distribution = [
750		- 'top_3' => 0,
751		- '4_10' => 0,
752		- '10_50' => 0,
753		- '51_100' => 0
754		- ];
	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	+ );
755	799
756		- foreach ($search_performance['rows'] ?? [] as $row) {
757		- $position = $row['position'] ?? 0;
758		- if ($position <= 3) {
759		- $position_distribution['top_3']++;
760		- } elseif ($position <= 10) {
761		- $position_distribution['4_10']++;
762		- } elseif ($position <= 50) {
763		- $position_distribution['10_50']++;
764		- } elseif ($position <= 100) {
765		- $position_distribution['51_100']++;
766		- }
767		- }
768		-
769	800	$dashboard_data['search_performance'] = array_merge($search_performance, [
770	801	'totals' => $totals,
771	802	'position_distribution' => $position_distribution
772	803	]);
		@@ -808,8 +839,132 @@
808	839	return $dashboard_data;
809	840	}
810	841
811	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	+ /**
812	967	* Get Core Web Vitals for the analytics dashboard, cached independently
813	968	* of the dashboard payload.
814	969	*
815	970	* Successful results are cached for 1 hour; failures are never cached
		@@ -852,9 +1007,9 @@
852	1007	* @param string $date_range Date range for analysis
853	1008	* @return array SEO opportunities
854	1009	*/
855	1010	public function get_seo_opportunities(string $date_range = '30d'): array {
856		- $cache_key = "seo_opportunities_{$date_range}";
	1011	+ $cache_key = self::OPPORTUNITIES_CACHE_PREFIX . $date_range;
857	1012	$cached_data = get_transient($cache_key);
858	1013
859	1014	if ($cached_data !== false) {
860	1015	return $cached_data;
		@@ -968,8 +1123,27 @@
968	1123	}
969	1124	}
970	1125
971	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	+ /**
972	1146	* Force refresh of all cached data
973	1147	*
974	1148	* @return array Refresh results
975	1149	*/
		@@ -976,21 +1150,12 @@
976	1150	public function refresh_data(): array {
977	1151	// Clear all analytics-related transients, including the previous-period
978	1152	// ranges used for trend comparison (14d/60d/180d) and the separately
979	1153	// cached Core Web Vitals payload.
980		- $cache_keys = [
981		- 'analytics_dashboard_v5_7d',
982		- 'analytics_dashboard_v5_30d',
983		- 'analytics_dashboard_v5_90d',
984		- 'analytics_dashboard_v5_14d',
985		- 'analytics_dashboard_v5_60d',
986		- 'analytics_dashboard_v5_180d',
987		- 'seo_opportunities_7d',
988		- 'seo_opportunities_30d',
989		- 'seo_opportunities_90d',
990		- 'indexing_status',
991		- 'thinkrank_dashboard_cwv'
992		- ];
	1154	+ $cache_keys = array_merge(
	1155	+ self::dashboard_cache_keys(),
	1156	+ ['indexing_status', 'thinkrank_dashboard_cwv']
	1157	+ );
993	1158
994	1159	// Also clear PageSpeed-derived caches. Their keys are md5-derived from
995	1160	// URL + device, so compute them for the URL/device combinations the
996	1161	// plugin actually tests.