Indeed. A big chunk of my email parser deals with missing or incorrect content headers. Most of the rest attempts to sensibly interpret the infinite combinations of parts found in multipart (and single-part!) emails.