Remove extraneous characters after .com

horgh · horgh · commit 4cf2c56517ef · 2024-01-26T12:26:09.000-08:00
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -17,6 +17,9 @@
 * Duplicate `.com`s are now removed from email domain names when
   `hash_address` is used. For example, `example.com.com` will become
   `example.com`.
+* Extraneous characters after `.com` are now removed from email domain
+  names when `hash_address` is used. For example, `example.comfoo` will
+  become `example.com`.
 
 ## v2.4.0 (2024-01-12)
 
diff --git a/lib/minfraud/components/email.rb b/lib/minfraud/components/email.rb
@@ -327,6 +327,7 @@ def clean_domain(domain)
         domain = SimpleIDN.to_ascii(domain)
 
         domain.sub!(/(?:\.com){2,}$/, '.com')
+        domain.sub!(/\.com[^.]+$/, '.com')
 
         if TYPO_DOMAINS.key?(domain)
           domain = TYPO_DOMAINS[domain]
diff --git a/spec/components/email_spec.rb b/spec/components/email_spec.rb
@@ -63,7 +63,7 @@
         { input: '  Test@maxmind.com', output: 'test@maxmind.com' },
         {
           input:  'Test@maxmind.com|abc124472372',
-          output: 'test@maxmind.com|abc124472372',
+          output: 'test@maxmind.com',
         },
         { input: 'Test+foo@yahoo.com', output: 'test+foo@yahoo.com' },
         { input: 'Test-foo@yahoo.com', output: 'test@yahoo.com' },
@@ -76,6 +76,7 @@
         { input: 'alias@user.fastmail.com', output: 'user@fastmail.com' },
         { input: 'foo-bar@ymail.com', output: 'foo@ymail.com' },
         { input: 'foo@example.com.com', output: 'foo@example.com' },
+        { input: 'foo@example.comfoo', output: 'foo@example.com' },
       ]
 
       tests.each do |i|

-Original file line number
+Diff line change
         { input: '  [email protected]', output: '[email protected]' },
+        {
           input:  '[email protected]|abc124472372',
 -          output: '[email protected]|abc124472372',
 +          output: '[email protected]',
         },
         { input: '[email protected]', output: '[email protected]' },
         { input: '[email protected]', output: '[email protected]' },
         { input: '[email protected]', output: '[email protected]' },
         { input: '[email protected]', output: '[email protected]' },
         { input: '[email protected]', output: '[email protected]' },
 +        { input: '[email protected]', output: '[email protected]' },
+      ]
       tests.each do |i|