fix(wikipedia): handle branded TLDs in extractBrandFromUrl

hdfc.bank.in was resolving to companyName=bank instead of hdfc because
.bank was not recognized as a multi-part TLD prefix. Add bank, firm,
gen, ind, res, nic to the prefix set (India ccSLD patterns).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: HollywoodTonight

src/wikipedia-analysis/handler.js

@@ -22,6 +22,7 @@ const REGION_SUFFIXES_RE = /(?:usa|us|uk|eu|de|fr|es|it|nl|be|at|ch|au|ca|jp|kr|
 
 const MULTI_PART_TLD_PREFIXES = new Set([
   'co', 'com', 'org', 'net', 'ac', 'gov', 'edu', 'mil',
+  'bank', 'firm', 'gen', 'ind', 'res', 'nic',
 ]);
 
 /**

test/audits/wikipedia-analysis/handler.test.js

@@ -807,5 +807,16 @@ describe('Wikipedia Analysis Handler', () => {
       expect(extractBrandFromUrl('https://a.b.walmart.com')).to.equal('walmart');
       expect(extractBrandFromUrl('https://dev.blog.google.com')).to.equal('google');
     });
+
+    it('should handle branded TLDs like .bank.in', () => {
+      expect(extractBrandFromUrl('https://hdfc.bank.in')).to.equal('hdfc');
+      expect(extractBrandFromUrl('https://icici.bank.in')).to.equal('icici');
+    });
+
+    it('should handle other industry-specific ccSLDs', () => {
+      expect(extractBrandFromUrl('https://acme.firm.in')).to.equal('acme');
+      expect(extractBrandFromUrl('https://iitd.res.in')).to.equal('iitd');
+      expect(extractBrandFromUrl('https://example.nic.in')).to.equal('example');
+    });
   });
 });