Sounds like a good idea to me. As for files not following the spec, the same can be said of HTML files and there are HTML parser modules available.

OT -- Ok, am I the only one to find it ironic that a format that is used for books was authored by someone named Thomas A. Sawyer, whose name is very similar to one of the most well known characters from Mark Twain's books (Tom Sawyer)? Sorry, couldn't resist commenting on the irony of that.