>" | "@"
                        | "," | ";" | ":" | "\" | <">
                        | "/" | "[" | "]" | "?" | "="
                        | "{" | "}" | SP | HT

      quoted-string     = ( <"> *(qdtext | quoted-pair ) <"> )
      qdtext            = <any TEXT except <">>
      quoted-pair       = "\" CHAR

      parameter         = attribute "=" value
      attribute         = token
      value             = token | quoted-string

    Each header is represented by a list of key/value pairs.  The value for a
    simple token (not part of a parameter) is None.  Syntactically incorrect
    headers will not necessarily be parsed as you would want.

    This is easier to describe with some examples:

    >>> split_header_words(['foo="bar"; port="80,81"; discard, bar=baz'])
    [[('foo', 'bar'), ('port', '80,81'), ('discard', None)], [('bar', 'baz')]]
    >>> split_header_words(['text/html; charset="iso-8859-1"'])
    [[('text/html', None), ('charset', 'iso-8859-1')]]
    >>> split_header_words([r'Basic realm="\"foo\bar\""'])
    [[('Basic', None), ('realm', '"foobar"')]]

    r