← All changes | includes/api/class-site-identity-endpoint.php +152 -19 1.28.0 → 2.9.0 View file →

		@@ -19,8 +19,9 @@
19	19
20	20	use ThinkRank\SEO\Site_Identity_Manager;
21	21	use ThinkRank\AI\Manager as AI_Manager;
22	22	use ThinkRank\API\Traits\CSRF_Protection;
	23	+use ThinkRank\API\Traits\Context_Authorization;
23	24	use WP_REST_Controller;
24	25	use WP_REST_Request;
25	26	use WP_REST_Response;
26	27	use WP_Error;
		@@ -31,8 +32,9 @@
31	32	}
32	33
33	34	// Load CSRF Protection trait
34	35	require_once THINKRANK_PLUGIN_DIR . 'includes/api/traits/trait-csrf-protection.php';
	36	+require_once THINKRANK_PLUGIN_DIR . 'includes/api/traits/trait-context-authorization.php';
35	37
36	38	/**
37	39	* Site Identity API Endpoints Class
38	40	*
		@@ -44,8 +46,9 @@
44	46	* @since 1.0.0
45	47	*/
46	48	class Site_Identity_Endpoint extends WP_REST_Controller {
47	49	use CSRF_Protection;
	50	+ use Context_Authorization;
48	51
49	52	/**
50	53	* Site Identity Manager instance
51	54	*
		@@ -126,9 +129,10 @@
126	129	[
127	130	[
128	131	'methods' => 'GET',
129	132	'callback' => [$this, 'get_settings'],
130		- 'permission_callback' => [$this, 'check_permissions']
	133	+ 'permission_callback' => [$this, 'check_permissions'],
	134	+ 'args' => $this->get_context_route_args()
131	135	],
132	136	[
133	137	'methods' => 'POST',
134	138	'callback' => [$this, 'update_settings'],
		@@ -278,14 +282,19 @@
278	282	*
279	283	* @since 1.0.0
280	284	*
281	285	* @param WP_REST_Request $request Request object
282		- * @return WP_REST_Response Response object
	286	+ * @return WP_REST_Response|WP_Error Response object, or the context error
283	287	*/
284		- public function get_settings(WP_REST_Request $request): WP_REST_Response {
	288	+ public function get_settings(WP_REST_Request $request) {
285	289	try {
286		- $context_type = $request->get_param('context_type') ?? 'site';
287		- $context_id = $request->get_param('context_id');
	290	+ // SECURITY: the settings are stored per context, so the object has
	291	+ // to be authorised before it is read (#385).
	292	+ $context = $this->resolve_request_context($request);
	293	+ if (is_wp_error($context)) {
	294	+ return $context;
	295	+ }
	296	+ [$context_type, $context_id] = $context;
288	297
289	298	// Get settings from Site Identity Manager
290	299	$settings = $this->identity_manager->get_settings($context_type, $context_id);
291	300
		@@ -321,11 +330,17 @@
321	330	*/
322	331	public function update_settings(WP_REST_Request $request) {
323	332	try {
324	333	$settings = $request->get_param('settings');
325		- $context_type = $request->get_param('context_type') ?? 'site';
326		- $context_id = $request->get_param('context_id');
327	334
	335	+ // SECURITY: this write is keyed by the context, so the object has to
	336	+ // be authorised before anything is persisted (#385).
	337	+ $context = $this->resolve_request_context($request);
	338	+ if (is_wp_error($context)) {
	339	+ return $context;
	340	+ }
	341	+ [$context_type, $context_id] = $context;
	342	+
328	343	// Validate settings
329	344	if (empty($settings) || !is_array($settings)) {
330	345	return new WP_Error(
331	346	'invalid_settings',
		@@ -354,10 +369,13 @@
354	369
355	370	if (!$update_result) {
356	371	return new WP_Error(
357	372	'update_failed',
358		- 'Failed to update site identity settings',
359		- ['status' => 500]
	373	+ $this->describe_save_failure('Failed to update site identity settings'),
	374	+ [
	375	+ 'status' => 500,
	376	+ 'failure_code' => $this->identity_manager->get_last_save_error_code()
	377	+ ]
360	378	);
361	379	}
362	380
363	381	// If this save changed the robots.txt content/toggle and a physical
		@@ -365,11 +383,17 @@
365	383	// static file directly (bypassing the robots_txt filter), so without
366	384	// this the file goes stale and /robots.txt shows the old content
367	385	// while the textarea shows the new — regardless of what the frontend
368	386	// believed about the file's existence.
	387	+ //
	388	+ // ai_crawler_rules counts as a robots.txt change even though it is
	389	+ // not the body: the directives it produces are composed into the
	390	+ // served output at render time, so a physical file left alone here
	391	+ // would keep serving the previous allow/block set (#657).
369	392	if ($context_type === 'site'
370	393	&& (array_key_exists('robots_txt_content', $settings)
371		- || array_key_exists('robots_txt_enabled', $settings))
	394	+ || array_key_exists('robots_txt_enabled', $settings)
	395	+ || array_key_exists('ai_crawler_rules', $settings))
372	396	) {
373	397	$this->identity_manager->sync_robots_txt_file();
374	398	}
375	399
		@@ -386,9 +410,9 @@
386	410	],
387	411	'message' => 'Site identity settings updated successfully'
388	412	], 200);
389	413
390		- } catch (\Exception $e) {
	414	+ } catch (\Throwable $e) {
391	415	return new WP_Error(
392	416	'update_failed',
393	417	'Settings update failed: ' . $e->getMessage(),
394	418	['status' => 500]
		@@ -537,8 +561,34 @@
537	561	// file if present, else the effective content) — header-stripped so
538	562	// the auto-generated comment/timestamp never lands in the textarea.
539	563	$robots_data['content'] = $this->identity_manager->get_served_robots_body();
540	564
	565	+ // Keep `rules` describing that same body. generate_robots_txt()
	566	+ // returned the rules it generated, which stopped matching `content`
	567	+ // the moment a stored override or a physical file supplied it.
	568	+ $robots_data['rules'] = $this->identity_manager->parse_robots_txt_rules($robots_data['content']);
	569	+
	570	+ // How /robots.txt is actually delivered right now, so the screen can
	571	+ // show the served output next to the editable body and flag a
	572	+ // physical file in the web root that has drifted from the settings.
	573	+ $robots_data['effective'] = $this->identity_manager->get_robots_txt_delivery();
	574	+
	575	+ // The per-agent AI crawler surface (#657). The registry ships with
	576	+ // the response rather than being duplicated in the bundle, so a
	577	+ // crawler added by the `thinkrank_ai_crawlers` filter appears in
	578	+ // the UI without a rebuild. Rules are returned normalised, so a
	579	+ // crawler with nothing stored comes back explicitly allowed rather
	580	+ // than as an absence the client has to interpret.
	581	+ $settings = $this->identity_manager->get_settings('site');
	582	+ $rules = \ThinkRank\SEO\AI_Crawlers::normalize_rules($settings['ai_crawler_rules'] ?? []);
	583	+
	584	+ $robots_data['ai_crawlers'] = \ThinkRank\SEO\AI_Crawlers::for_display();
	585	+ $robots_data['ai_crawler_rules'] = [];
	586	+
	587	+ foreach ($robots_data['ai_crawlers'] as $agent) {
	588	+ $robots_data['ai_crawler_rules'][$agent['slug']] = $rules[$agent['slug']] ?? 'allow';
	589	+ }
	590	+
541	591	return new WP_REST_Response([
542	592	'success' => true,
543	593	'data' => $robots_data,
544	594	'message' => 'Robots.txt data retrieved successfully'
		@@ -594,8 +644,17 @@
594	644	}
595	645
596	646	$settings = ['robots_txt_enabled' => $enable_management];
597	647
	648	+ // Per-agent AI crawler rules (#657). Only written when the caller
	649	+ // sends them: this route is also the plain "re-sync the file" save,
	650	+ // and defaulting a missing parameter to an empty map there would
	651	+ // unblock every crawler the site had blocked.
	652	+ $ai_rules = $request->get_param('ai_crawler_rules');
	653	+ if (null !== $ai_rules) {
	654	+ $settings['ai_crawler_rules'] = \ThinkRank\SEO\AI_Crawlers::normalize_rules($ai_rules);
	655	+ }
	656	+
598	657	if ($regenerate) {
599	658	// Generate and validate robots.txt from the rules.
600	659	$robots_data = $this->identity_manager->generate_robots_txt($custom_rules);
601	660
		@@ -619,10 +678,13 @@
619	678
620	679	if (!$update_result) {
621	680	return new WP_Error(
622	681	'update_failed',
623		- 'Failed to update robots.txt settings',
624		- ['status' => 500]
	682	+ $this->describe_save_failure('Failed to update robots.txt settings'),
	683	+ [
	684	+ 'status' => 500,
	685	+ 'failure_code' => $this->identity_manager->get_last_save_error_code()
	686	+ ]
625	687	);
626	688	}
627	689
628	690	// Effective content = the stored textarea content when set, else the
		@@ -645,9 +707,14 @@
645	707	$robots_data = $robots_data ?? [];
646	708	// Return the header-stripped body so the client textarea reflects
647	709	// exactly what it should hold (the header is added only at render).
648	710	$robots_data['content'] = $this->identity_manager->get_served_robots_body();
	711	+ $robots_data['rules'] = $this->identity_manager->parse_robots_txt_rules($robots_data['content']);
649	712
	713	+ // Re-read delivery after the write above so the screen reflects the
	714	+ // file that now exists rather than the state it was in on load.
	715	+ $robots_data['effective'] = $this->identity_manager->get_robots_txt_delivery();
	716	+
650	717	return new WP_REST_Response([
651	718	'success' => true,
652	719	'data' => [
653	720	'robots_data' => $robots_data,
		@@ -658,9 +725,9 @@
658	725	? 'Robots.txt configuration updated and file written successfully'
659	726	: 'Robots.txt configuration updated (file not written: ' . $file_write_result['message'] . ')'
660	727	], 200);
661	728
662		- } catch (\Exception $e) {
	729	+ } catch (\Throwable $e) {
663	730	return new WP_Error(
664	731	'update_failed',
665	732	'Robots.txt update failed: ' . $e->getMessage(),
666	733	['status' => 500]
		@@ -859,12 +926,14 @@
859	926	*/
860	927	public function validate_identity_settings(WP_REST_Request $request): WP_REST_Response {
861	928	try {
862	929	$settings = $request->get_param('settings');
863		- $tab_context = $request->get_param('tab_context') ?? '';
	930	+ $tab_context = $request->get_param('tab_context');
864	931
865		- // Validate settings using Site Identity Manager with tab context
866		- $validation = $this->identity_manager->validate_settings($settings ?? [], $tab_context);
	932	+ // Validate settings using Site Identity Manager with tab context.
	933	+ // `settings` is registered required, so REST rejects a missing value
	934	+ // before this point and the old `?? []` fallback was unreachable.
	935	+ $validation = $this->identity_manager->validate_settings($settings, $tab_context);
867	936
868	937	return new WP_REST_Response([
869	938	'success' => true,
870	939	'data' => $validation,
		@@ -923,8 +992,28 @@
923	992	* Argument validation methods
924	993	*/
925	994
926	995	/**
	996	+ * Build a save-failure message that names the actual cause.
	997	+ *
	998	+ * The manager knows why the save failed — missing settings table, rejected
	999	+ * INSERT with the MySQL error attached — and used to write that to the
	1000	+ * error log and throw it away, leaving the client a fixed string that told
	1001	+ * nobody anything. Append the reason so the response is diagnosable on its
	1002	+ * own. Status stays 500: a rejected INSERT is a server-side failure.
	1003	+ *
	1004	+ * @since 1.32.1
	1005	+ *
	1006	+ * @param string $fallback Message to use when no reason was recorded.
	1007	+ * @return string Failure message.
	1008	+ */
	1009	+ private function describe_save_failure(string $fallback): string {
	1010	+ $reason = $this->identity_manager->get_last_save_error();
	1011	+
	1012	+ return '' !== $reason ? $fallback . ': ' . $reason : $fallback;
	1013	+ }
	1014	+
	1015	+ /**
927	1016	* Get arguments for settings endpoints
928	1017	*
929	1018	* @since 1.0.0
930	1019	*
		@@ -964,9 +1053,12 @@
964	1053	return [
965	1054	'template_name' => [
966	1055	'required' => false,
967	1056	'type' => 'string',
968		- 'enum' => ['default', 'simple', 'reverse', 'category', 'author'],
	1057	+ // Must match get_available_title_templates(), which returns
	1058	+ // 'default' and nothing else. The extra names advertised
	1059	+ // templates the resolver has never been able to produce.
	1060	+ 'enum' => ['default'],
969	1061	'default' => 'default',
970	1062	'description' => 'Title template to use'
971	1063	],
972	1064	'data' => [
		@@ -995,9 +1087,11 @@
995	1087	return [
996	1088	'breadcrumb_type' => [
997	1089	'required' => false,
998	1090	'type' => 'string',
999		- 'enum' => ['hierarchical', 'taxonomy', 'path', 'custom'],
	1091	+ // Must match get_available_breadcrumb_types(), which returns
	1092	+ // 'hierarchical' and nothing else.
	1093	+ 'enum' => ['hierarchical'],
1000	1094	'default' => 'hierarchical',
1001	1095	'description' => 'Type of breadcrumb navigation to generate'
1002	1096	],
1003	1097	'options' => [
		@@ -1035,8 +1129,29 @@
1035	1129	'required' => false,
1036	1130	'type' => 'boolean',
1037	1131	'default' => true,
1038	1132	'description' => 'Rebuild content from rules (Generate). When false, only re-sync the physical file to the stored content.'
	1133	+ ],
	1134	+ // The handler reads this and the route never declared it, so it
	1135	+ // arrived as whatever string the client sent. RobotsManagement.js
	1136	+ // sends it, and "false" is a non-empty string — truthy — so the
	1137	+ // file was written when the caller had asked it not to be. ("0" is
	1138	+ // falsy, which is why the failure was asymmetric.) Registering it
	1139	+ // gets core's boolean coercion (#394).
	1140	+ 'ai_crawler_rules' => [
	1141	+ 'required' => false,
	1142	+ 'type' => 'object',
	1143	+ 'description' => 'Per-agent AI crawler rules, keyed by crawler slug, each "allow" or "block".',
	1144	+ 'additionalProperties' => [
	1145	+ 'type' => 'string',
	1146	+ 'enum' => ['allow', 'block'],
	1147	+ ],
	1148	+ ],
	1149	+ 'write_to_file' => [
	1150	+ 'required' => false,
	1151	+ 'type' => 'boolean',
	1152	+ 'default' => true,
	1153	+ 'description' => 'Write the generated content to the physical robots.txt file.'
1039	1154	]
1040	1155	];
1041	1156	}
1042	1157
		@@ -1052,8 +1167,16 @@
1052	1167	'identity_data' => [
1053	1168	'required' => true,
1054	1169	'type' => 'object',
1055	1170	'description' => 'Site identity data to optimize'
	1171	+ ],
	1172	+ // Read by optimize_site_identity(); previously unregistered, so it
	1173	+ // never appeared in the published schema.
	1174	+ 'options' => [
	1175	+ 'required' => false,
	1176	+ 'type' => 'object',
	1177	+ 'default' => [],
	1178	+ 'description' => 'Additional optimization options'
1056	1179	]
1057	1180	];
1058	1181	}
1059	1182
		@@ -1200,8 +1323,18 @@
1200	1323	'settings' => [
1201	1324	'required' => true,
1202	1325	'type' => 'object',
1203	1326	'description' => 'Settings to validate'
	1327	+ ],
	1328	+ // Read by validate_identity_settings() to scope validation to one
	1329	+ // tab; it was never registered, so it was absent from the published
	1330	+ // schema and got no type or sanitization.
	1331	+ 'tab_context' => [
	1332	+ 'required' => false,
	1333	+ 'type' => 'string',
	1334	+ 'default' => '',
	1335	+ 'sanitize_callback' => 'sanitize_key',
	1336	+ 'description' => 'Limit validation to a single settings tab'
1204	1337	]
1205	1338	];
1206	1339	}
1207	1340