Find duplicated code across large code bases. Works only on C and C++ for now, but could in principle work for many other languages, even when repeated code is rewritten between different languages. Unpaste already "sees through" common minor changes that may happen between copies of the same code over time.

The algorithm strips code down to an internal minimised format, leaving out most of the details. You couldn't reconstruct a working program from this format, but it can be helpful in spotting identical or near-identical code that's been copied around.