I found a PDF file that begin with the sequence:
firmas_t/1082774737.png%PDF-1.7
So, the method filetype.match returns None. According to this stackoverflow answer, some programs can generate this, but are still valid PDF files. It's possible to extend support for this files ?