Normalize equivalent domain names

Author: horgh

CHANGELOG.md

@@ -1,5 +1,10 @@
 # Changelog
 
+## v2.5.0
+
+* Equivalent domain names are now normalized when `hash_address` is used.
+  For example, `googlemail.com` will become `gmail.com`.
+
 ## v2.4.0 (2024-01-12)
 
 * Ruby 2.7+ is now required. If you're using Ruby 2.5 or 2.6, please use

lib/minfraud/components/email.rb

@@ -113,6 +113,18 @@ def clean_email_address(address)
       }.freeze
       private_constant :TYPO_DOMAINS
 
+      EQUIVALENT_DOMAINS = {
+        'googlemail.com' => 'gmail.com',
+        'pm.me'          => 'protonmail.com',
+        'proton.me'      => 'protonmail.com',
+        'yandex.by'      => 'yandex.ru',
+        'yandex.com'     => 'yandex.ru',
+        'yandex.kz'      => 'yandex.ru',
+        'yandex.ua'      => 'yandex.ru',
+        'ya.ru'          => 'yandex.ru',
+      }.freeze
+      private_constant :EQUIVALENT_DOMAINS
+
       def clean_domain(domain)
         domain = domain.strip
 
@@ -125,6 +137,10 @@ def clean_domain(domain)
           domain = TYPO_DOMAINS[domain]
         end
 
+        if EQUIVALENT_DOMAINS.key?(domain)
+          domain = EQUIVALENT_DOMAINS[domain]
+        end
+
         domain
       end
     end

spec/components/email_spec.rb

@@ -71,6 +71,7 @@
         { input: '[email protected]', output: '[email protected]' },
         { input: '[email protected]', output: '[email protected]' },
         { input: 'Test+alias@bücher.com', output: '[email protected]' },
+        { input: '[email protected]', output: '[email protected]' },
       ]
 
       tests.each do |i|