Set the result encoding of CGI.encode to UTF-8.

Author: gkellogg

lib/rdf/model/uri.rb

@@ -902,7 +902,7 @@ def user=(value)
     # Normalized version of user
     # @return [String]
     def normalized_user
-      URI.encode(CGI.unescape(user), /[^#{IUNRESERVED}|#{SUB_DELIMS}]/) if user
+      URI.encode(CGI.unescape(user), /[^#{IUNRESERVED}|#{SUB_DELIMS}]/).encode(Encoding::UTF_8) if user
     end
 
     ##
@@ -928,7 +928,7 @@ def password=(value)
     # Normalized version of password
     # @return [String]
     def normalized_password
-      URI.encode(CGI.unescape(password), /[^#{IUNRESERVED}|#{SUB_DELIMS}]/) if password
+      URI.encode(CGI.unescape(password), /[^#{IUNRESERVED}|#{SUB_DELIMS}]/).encode(Encoding::UTF_8) if password
     end
 
     HOST_FROM_AUTHORITY_RE = /(?:[^@]+@)?([^:]+)(?::.*)?$/.freeze
@@ -1295,7 +1295,7 @@ def normalize_segment(value, expr, downcase = false)
         value = value.dup.force_encoding(Encoding::UTF_8)
         decoded = CGI.unescape(value)
         decoded.downcase! if downcase
-        URI.encode(decoded, /[^(?:#{expr})]/)
+        URI.encode(decoded, /[^(?:#{expr})]/).encode(Encoding::UTF_8)
       end
     end
 

spec/model_uri_spec.rb

@@ -486,11 +486,6 @@
       expect(u1.canonicalize!.to_s).to eq u2.to_s
       expect(u1).to eq u2
     end
-    it "#canonicalize does not fail with Encoding::CompatibilityError on weird IRIs" do
-      u1 = RDF::URI "htЫtp://user:passoЫd@exaЫmple.com:8080/path ПУТЬ?queЫry=valЫue#fragmeЫnt"
-      u2 = RDF::URI "ht%D0%ABtp://user:passoЫd@exaЫmple.com:8080/path%20ПУТЬ?queЫry=valЫue#fragmeЫnt"
-      expect {u1.canonicalize.to_s.dup.force_encoding(u2.to_s.encoding)}.not_to raise_error
-    end
   end
 
   describe "#/" do