Should not assume string.lower().upper() == string.upper()

[petterreinholdtsen]

The code in header_canonical() transform headers first to lower case and then to upper case, while the referred text in https://docs.djangoproject.com/en/1.7/ref/request-response/#django.http.HttpRequest.META (1.6 do no longer exist) state:

With the exception of CONTENT_LENGTH and CONTENT_TYPE, as given above, any HTTP headers in the request are converted to META keys by converting all characters to uppercase, replacing any hyphens with underscores and adding an HTTP_ prefix to the name.

It is not safe to assume that, first converting to lower case and then to upper case will result in the same string as going directly to upper case. For example, depending on the locale, the Turkish dottet capital I, can result in a dodless lower case I, which in turn end up with a dotless capital I. See https://en.wikipedia.org/wiki/Turkish_dotted_and_dotless_I for more info on this example. I am sure there are other examples.

I recommend going directly to upper case in the code.