I don’t know what you expected. File is just there to give a good guess at a file’s format. There are a ton of reasons why this problem is hard, and there are reasons to make “file” less accurate in order to make the implementation simpler and more secure.

But it will work fine for PDFs, often enough.