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}]/).encode(Encoding::UTF_8) if user
+      URI.encode(CGI.unescape(user), /[^#{IUNRESERVED}|#{SUB_DELIMS}]/).force_encoding(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}]/).encode(Encoding::UTF_8) if password
+      URI.encode(CGI.unescape(password), /[^#{IUNRESERVED}|#{SUB_DELIMS}]/).force_encoding(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})]/).encode(Encoding::UTF_8)
+        URI.encode(decoded, /[^(?:#{expr})]/).force_encoding(Encoding::UTF_8)
       end
     end
 

spec/model_uri_spec.rb

@@ -470,14 +470,18 @@
       "urn" => [
         "urn:ex:s001",
         "urn:ex:s001"
+      ],
+      "utf-8" => [
+        "Dürst",
+        "Dürst"
       ]
     }.each do |name, (input, output)|
       it "#canonicalize #{name}" do
         u1 = RDF::URI(input)
         u2 = RDF::URI(output)
         expect(u1.canonicalize.hash).to eq u2.hash
         expect(u1.canonicalize.to_s).to eq u2.to_s
-        expect(u1).to eq u1
+        expect(u1.canonicalize).to eq u1.canonicalize
       end
     end
     it "#canonicalize! alters resource" do