encoding.rb in contrast-agent-7.5.0

- old
+ new

@@ -35,11 +35,11 @@
             # Making an exception list is not a good idea, because it will be hard to maintain.
             #
             # The Base64 method will return printable ascii characters, so we can use this to determine if the value is
             # encoded or not.
             #
-            # The solution in this case is encodind the value, and then decoding it. If the value is already encoded
+            # The solution in this case is encoding the value, and then decoding it. If the value is already encoded
             # it will not be eq to the original value. If the value is not encoded, it will be eq to the original value.
             #
             # @param value [String] input to check for encoding status.
             # @param input_type [Symbol] input type.
             # @return [Boolean] true if value is base64 encoded, false otherwise.
@@ -94,13 +94,45 @@
             # @return [String] decoded or original value.
             # @raise [StandardError]
             def cs__decode64 value, input_type
               return value unless cs__base64?(value, input_type)
 
-              Base64.decode64(value)
+              new_value = try_base64_decode(value)
+              new_value, success = normalize_encoding(new_value)
+              return new_value if success
+
+              value
             rescue StandardError => e
               logger.error('Error while decoding base64', error: e, message: e.message, backtrace: e.backtrace)
               value
+            end
+
+            private
+
+            # Try and decode the value, do not use decoding if the value have zero bytes.
+            def try_base64_decode value
+              new_value = Base64.decode64(value)
+              # check for null bytes:
+              return new_value unless new_value.bytes.select(&:zero?).any?
+
+              value
+            end
+
+            # Detecting encoded Base64 is not perfect. In some cases it will detect certain inputs as
+            # encoded and will try to decode them. Even if the decoding is successful, the value may be
+            # encoded back to ASCII format. AgentLib will raise UTF-8 error in this case.
+            # This method will try to normalize the encoding to UTF-8. If the encoding fails, this means
+            # that the decoding was not successful and the value will be returned as is. Otherwise a
+            # base64 decoded string with ASCII-8BIT encoding will be parsed to UTF-8 without errors.
+            #
+            # @param value [String] input to normalize.
+            # @return [Array<String, Boolean>] normalized value and success flag.
+            def normalize_encoding value
+              new_value = value.dup.encode!('Windows-1252').force_encoding('UTF-8')
+              [new_value, true]
+            rescue StandardError
+              # encoding failed, or the decoding from base64 failed.
+              [nil, false]
             end
           end
         end
       end
     end