Validate domain hostname rules per label

lib/valid_email2/address.rb

@@ -11,6 +11,9 @@ class Address
     PROHIBITED_DOMAIN_CHARACTERS_REGEX = /[+!_\/\s'#`]/
     DEFAULT_RECIPIENT_DELIMITER = '+'
     DOT_DELIMITER = '.'
+    HYPHEN_DELIMITER = '-'
+    MAX_DOMAIN_LENGTH = 253
+    MAX_LABEL_LENGTH = 63
 
     def self.prohibited_domain_characters_regex
       @prohibited_domain_characters_regex ||= PROHIBITED_DOMAIN_CHARACTERS_REGEX
@@ -52,13 +55,13 @@ def valid?
     def valid_domain?
       domain = address.domain
       return false if domain.nil?
+      return false if domain =~ self.class.prohibited_domain_characters_regex
+      return false if domain.length > MAX_DOMAIN_LENGTH
 
-      domain !~ self.class.prohibited_domain_characters_regex &&
-        domain.include?('.') &&
-        !domain.include?('..') &&
-        !domain.start_with?('.') &&
-        !domain.start_with?('-') &&
-        !domain.include?('-.')
+      labels = domain.split(DOT_DELIMITER, -1)
+      return false if labels.length < 2
+
+      labels.all? { |label| valid_domain_label?(label) }
     end
 
     def valid_address?
@@ -109,6 +112,14 @@ def valid_strict_mx?
 
     private
 
+    # An RFC 1035 label: 1-63 octets, and it may not begin or end with a hyphen.
+    def valid_domain_label?(label)
+      !label.empty? &&
+        label.length <= MAX_LABEL_LENGTH &&
+        !label.start_with?(HYPHEN_DELIMITER) &&
+        !label.end_with?(HYPHEN_DELIMITER)
+    end
+
     def disposable_mx_server?
       mx_server_is_in?(ValidEmail2.disposable_emails)
     end

spec/valid_email2_spec.rb

@@ -171,6 +171,48 @@ class TestUserMultiple < TestModel
       user = TestUser.new(email: "foo@example.com#")
       expect(user.valid?).to be_falsey
     end
+
+    describe "domain label structure" do
+      # A domain label may not begin or end with a hyphen and is limited to 63
+      # octets; the whole domain is limited to 253. These rules apply to every
+      # label, not only the first/last position of the domain string.
+      long_label = "a" * 64
+      long_domain = (["aa"] + ["a"] * 126).join(".") # 254 chars, all-valid labels
+
+      {
+        "a leading hyphen on the first label" => "foo@-example.com",
+        "a leading hyphen on a later label" => "foo@sub.-example.com",
+        "a leading hyphen on a deeply nested label" => "foo@a.b.-c.example.com",
+        "a trailing hyphen on a non-final label" => "foo@example-.com",
+        "a trailing hyphen on the final label" => "foo@example.com-",
+        "a bare hyphen label" => "foo@example.-.com",
+        "a label longer than 63 octets" => "foo@#{long_label}.com",
+        "a later label longer than 63 octets" => "foo@example.#{long_label}.com",
+        "a domain longer than 253 octets" => "foo@#{long_domain}"
+      }.each do |description, email|
+        it "is invalid with #{description}" do
+          expect(TestUser.new(email: email).valid?).to be_falsey
+        end
+      end
+
+      full_label = "a" * 63
+      max_domain = (["a"] * 127).join(".") # exactly 253 chars
+
+      {
+        "an internal hyphen" => "foo@ex-ample.com",
+        "hyphens on several internal labels" => "foo@f-o-o.example.com",
+        "a double internal hyphen" => "foo@a--b.example.com",
+        "a punycode (IDNA ACE) label" => "foo@xn--bcher-kva.com",
+        "a label beginning with a digit" => "foo@3m.com",
+        "a 63 octet label at the limit" => "foo@#{full_label}.com",
+        "a later 63 octet label" => "foo@example.#{full_label}.com",
+        "a domain at the 253 octet limit" => "foo@#{max_domain}"
+      }.each do |description, email|
+        it "is valid with #{description}" do
+          expect(TestUser.new(email: email).valid?).to be_truthy
+        end
+      end
+    end
   end
 
   describe "with disposable validation" do