Currently thunk prefixes are mutally exclusive. No thunks database path
can currently have more than one prefix. If this is necessary then we
can add it in the future.
Most minor of optimizations here, we only scan the string once to find a
prefix, instead of searching for it on all four prefix replacements.