You can't. With human language, the wordspace is so large that plagiarism is often identifiable, but idiom and such are very common, so even there it's very har

You can't. With human language, the wordspace is so large that plagiarism is often identifiable, but idiom and such are very common, so even there it's very hard to tell derivative from original (and that's not even going into patterns and archetypes). With programming languages, the wordspace is usually tiny, the idioms are very many, and most programs closely follow an archetypical pattern. If whole sections are copied, you can't tell, because almost everyone uses those same operations in the same way, the same variable naming patterns, and the same formatting style. Lazy copiers might be easy to spot if there were inconsistencies in formatting throughout a program, comments that match copied-from-source, lots of extraneous glue that indicates lacking the understanding to factor and adapt techniques.