You'll never know for sure if a string is URL-encoded or if it was supposed to have the sequence %2B
in it. Instead, it probably depends on where the string came from, i.e. if it was hand-crafted or from some application.
Is it better to search the string for characters which would be encoded, which aren't, and if any exist then its not encoded.
I think this is a better approach, since it would take care of things that have been done programmatically (assuming the application would not have left a non-encoded character behind).
One thing that will be confusing here... Technically, the %
"should be" encoded if it will be present in the final value, since it is a special character. You might have to combine your approaches to look for should-be-encoded characters as well as validating that the string decodes successfully if none are found.