fix(LLMO-4176): skip brandalf check when onboardingMode is v1 in llmo-customer-analysis

src/llmo-customer-analysis/handler.js

@@ -321,33 +321,43 @@ export async function runLlmoCustomerAnalysis(finalUrl, context, site, auditCont
 
   const triggeredSteps = [];
   const hasOptelData = await checkOptelData(domain, context);
-  const { configVersion, previousConfigVersion } = auditContext;
+  const { configVersion, previousConfigVersion, onboardingMode } = auditContext;
   const isFirstTimeOnboarding = !previousConfigVersion;
 
-  // For brandalf-enabled orgs, resolve brand ID so the DRS scheduler can use v2 prompts
+  // For brandalf-enabled orgs, resolve brand ID so the DRS scheduler can use v2 prompts.
+  // If onboardingMode is explicitly 'v1' (set by api-service for mixed-state orgs with
+  // pre-Brandalf sites), skip v2 brand resolution — the org has brandalf=true but was
+  // onboarded via the v1 path, so no customer config brand exists yet.
   let brandId;
   let organizationId;
   if (isFirstTimeOnboarding) {
     const orgId = site.getOrganizationId?.() || auditContext.imsOrgId;
     if (orgId) {
-      const isV2 = await isBrandalfEnabled(orgId, env, log);
+      const isV2 = onboardingMode !== 'v1' && await isBrandalfEnabled(orgId, env, log);
       if (isV2) {
         organizationId = orgId;
         try {
           const { postgrestClient } = context.dataAccess?.services || {};
           if (postgrestClient?.from) {
+            // Prefer the brand whose baseSiteId (site_id column) matches this
+            // site — this is the primary brand created during onboarding with
+            // the correct base URL.  Fall back to brand_sites join if no
+            // baseSiteId match exists (backward compat for brands created
+            // before baseSiteId was set during onboarding).
             const { data: brands } = await postgrestClient
               .from('brands')
-              .select('id, brand_sites(site_id)')
+              .select('id, site_id, brand_sites(site_id)')
               .eq('organization_id', organizationId)
               .eq('status', 'active');
 
-            const match = brands?.find(
+            const baseSiteMatch = brands?.find((b) => b.site_id === siteId);
+            const brandSiteMatch = !baseSiteMatch && brands?.find(
               (b) => b.brand_sites?.some((bs) => bs.site_id === siteId),
             );
+            const match = baseSiteMatch || brandSiteMatch;
             if (match) {
               brandId = match.id;
-              log.info(`Resolved brand ${brandId} for site ${siteId} (v2 onboarding)`);
+              log.info(`Resolved brand ${brandId} for site ${siteId} (v2 onboarding, via ${baseSiteMatch ? 'baseSiteId' : 'brand_sites'})`);
             } else {
               log.warn(`No brand found matching site ${siteId} in org ${organizationId} for v2 BP schedule`);
             }

test/audits/llmo-customer-analysis.test.js

@@ -1587,7 +1587,7 @@ describe('LLMO Customer Analysis Handler', () => {
         select: sandbox.stub().returns({
           eq: sandbox.stub().returns({
             eq: sandbox.stub().resolves({
-              data: [{ id: 'brand-uuid-1', brand_sites: [{ site_id: 'site-123' }] }],
+              data: [{ id: 'brand-uuid-1', site_id: 'site-123', brand_sites: [{ site_id: 'site-123' }] }],
             }),
           }),
         }),
@@ -1652,7 +1652,7 @@ describe('LLMO Customer Analysis Handler', () => {
         select: sandbox.stub().returns({
           eq: sandbox.stub().returns({
             eq: sandbox.stub().resolves({
-              data: [{ id: 'brand-fb', brand_sites: [{ site_id: 'site-123' }] }],
+              data: [{ id: 'brand-fb', site_id: null, brand_sites: [{ site_id: 'site-123' }] }],
             }),
           }),
         }),
@@ -1694,7 +1694,7 @@ describe('LLMO Customer Analysis Handler', () => {
         select: sandbox.stub().returns({
           eq: sandbox.stub().returns({
             eq: sandbox.stub().resolves({
-              data: [{ id: 'brand-no-sites' }],
+              data: [{ id: 'brand-no-sites', site_id: null }],
             }),
           }),
         }),
@@ -1734,7 +1734,7 @@ describe('LLMO Customer Analysis Handler', () => {
         select: sandbox.stub().returns({
           eq: sandbox.stub().returns({
             eq: sandbox.stub().resolves({
-              data: [{ id: 'brand-other', brand_sites: [{ site_id: 'other-site' }] }],
+              data: [{ id: 'brand-other', site_id: 'other-site', brand_sites: [{ site_id: 'other-site' }] }],
             }),
           }),
         }),
@@ -1757,6 +1757,63 @@ describe('LLMO Customer Analysis Handler', () => {
       expect(log.warn).to.have.been.calledWith(sinon.match(/No brand found matching site/));
     });
 
+    it('should prefer baseSiteId match over brand_sites match', async () => {
+      const auditContext = {};
+
+      context.env.SPACECAT_API_BASE_URL = 'https://spacecat.example.com';
+      context.env.SPACECAT_API_KEY = 'test-api-key';
+
+      mockFetch.reset();
+      mockFetch.onFirstCall().resolves({
+        ok: true,
+        json: async () => [{ flagName: 'brandalf', flagValue: true }],
+      });
+      mockFetch.onSecondCall().resolves({
+        ok: true,
+        json: async () => ({ schedule_id: 'sched-001' }),
+      });
+      mockFetch.onThirdCall().resolves({
+        ok: true,
+        json: async () => ({}),
+      });
+
+      // Two brands: one with baseSiteId (site_id) match, one with only brand_sites match.
+      // The baseSiteId match should be preferred (it has the correct base URL).
+      const brandsQuery = {
+        select: sandbox.stub().returns({
+          eq: sandbox.stub().returns({
+            eq: sandbox.stub().resolves({
+              data: [
+                { id: 'brand-sub', site_id: null, brand_sites: [{ site_id: 'site-123' }] },
+                { id: 'brand-base', site_id: 'site-123', brand_sites: [{ site_id: 'site-123' }] },
+              ],
+            }),
+          }),
+        }),
+      };
+      context.dataAccess.services = {
+        postgrestClient: { from: sandbox.stub().returns(brandsQuery) },
+      };
+
+      mockLlmoConfig.readConfig.resolves({
+        config: {
+          entities: {},
+          categories: {},
+          topics: {},
+          brands: { aliases: [] },
+          competitors: { competitors: [] },
+        },
+      });
+
+      await mockHandler.runLlmoCustomerAnalysis('https://example.com', context, site, auditContext);
+
+      // Should pick brand-base (baseSiteId match) over brand-sub (brand_sites only)
+      const createCall = mockFetch.getCalls().find((c) => c.args[0] === 'https://drs.example.com/api/schedules');
+      expect(createCall).to.exist;
+      const body = JSON.parse(createCall.args[1].body);
+      expect(body.brand_id).to.equal('brand-base');
+    });
+
     it('should create BP schedule without brand_id when brand lookup fails', async () => {
       const auditContext = {};
 
@@ -1987,6 +2044,57 @@ describe('LLMO Customer Analysis Handler', () => {
       expect(body.brand_id).to.be.undefined;
     });
 
+    it('should skip brandalf check and omit brand_id when onboardingMode is v1 (mixed-state org)', async () => {
+      // Mixed-state: org has brandalf=true but was onboarded via v1 path because it has
+      // pre-Brandalf sites. onboardingMode='v1' in auditContext must bypass isBrandalfEnabled
+      // so we don't send brand_id to DRS (no customer config brand exists for v1 onboarding).
+      const auditContext = { onboardingMode: 'v1' };
+
+      context.env.SPACECAT_API_BASE_URL = 'https://spacecat.example.com';
+      context.env.SPACECAT_API_KEY = 'test-api-key';
+
+      mockFetch.reset();
+      // Only DRS calls expected — feature-flags API must NOT be called
+      mockFetch.onFirstCall().resolves({
+        ok: true,
+        json: async () => ({ schedule_id: 'sched-v1' }),
+      });
+      mockFetch.onSecondCall().resolves({
+        ok: true,
+        json: async () => ({}),
+      });
+
+      mockLlmoConfig.readConfig.resolves({
+        config: {
+          entities: {},
+          categories: {},
+          topics: {},
+          brands: { aliases: [] },
+          competitors: { competitors: [] },
+        },
+      });
+
+      await mockHandler.runLlmoCustomerAnalysis(
+        'https://example.com',
+        context,
+        site,
+        auditContext,
+      );
+
+      // feature-flags API must NOT have been called
+      const featureFlagsCall = mockFetch.getCalls().find(
+        (c) => typeof c.args[0] === 'string' && c.args[0].includes('/feature-flags'),
+      );
+      expect(featureFlagsCall).to.not.exist;
+
+      // Schedule should be created without brand_id
+      const scheduleCall = mockFetch.getCalls().find((c) => c.args[0] === 'https://drs.example.com/api/schedules');
+      expect(scheduleCall).to.exist;
+      const scheduleBody = JSON.parse(scheduleCall.args[1].body);
+      expect(scheduleBody.brand_id).to.be.undefined;
+      expect(scheduleBody.spacecat_org_id).to.be.undefined;
+    });
+
   });
 
 });