Comment 0 for bug 266574

This patch fixes a number of problems for non-English
archives. In particular:

- it normalizes the encoding name (converting to
lower-case, splitting of potential \r characters, and
splitting off potential quotes)

- if the article and the header encoding differ, it tries
to preserve the decoded header as a Unicode string,
instead of immediately falling back to the MIME-encoded
one (if the encoding is not known to Python it will
still fall back)

- if a Unicode string needs to be formatted, it uses
character references for all non-ASCII characters.

[http://sourceforge.net/tracker/index.php?func=detail&aid=510415&group_id=103&atid=300103]