I don't really see why there is a Content-Length in the headers of a
multipart form data. The specification at
http://www.w3.org/TR/html401/interact/forms.html#h-17.13.4.2 doesn't
mention it, and it is absent in the example that looks like the one tested
by Peter :

Content-Type: multipart/form-data; boundary=AaB03x
--AaB03x
Content-Disposition: form-data; name="submit-name"

Larry
--AaB03x
Content-Disposition: form-data; name="files"; filename="file1.txt"
Content-Type: text/plain
... contents of file1.txt ...
--AaB03x--

In case a user agent would insert it, I think the best would be to
ignore it. That is, inside read_multi(), after

            headers = parser.close()

add these lines :

            if 'content-length' in headers:
                del headers['content-length']

It's hard to see the potential side effects but I think it's cleaner
than the proposed patch, which is not correct anyway for another
reason : the attribute value is set to a bytes objects, instead of a
string.

Peter, does this make sense ? If so, can you submit another patch ?