[PATCH 07/15] Encode utf-8 argument for md5 function

[Nicolas Sebrecht]

On Tue, May 10, 2016 at 01:18:29AM +0200, Łukasz Żarnowiecki wrote:
> Python3 accepts binary input for md5 function.
> 
> Signed-off-by: Łukasz Żarnowiecki <dolohow at outlook.com>
> ---
>  offlineimap/folder/Maildir.py | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/offlineimap/folder/Maildir.py b/offlineimap/folder/Maildir.py
> index bda452a..95e2553 100644
> --- a/offlineimap/folder/Maildir.py
> +++ b/offlineimap/folder/Maildir.py
> @@ -71,7 +71,7 @@ class MaildirFolder(BaseFolder):
>          # Everything up to the first comma or colon (or ! if Windows):
>          self.re_prefixmatch = re.compile('([^'+ self.infosep + ',]*)')
>          # folder's md, so we can match with recorded file md5 for validity.
> -        self._foldermd5 = md5(self.getvisiblename()).hexdigest()
> +        self._foldermd5 = md5(self.getvisiblename().encode('utf-8')).hexdigest()

I'm not sure about that. I know users out there are using tricks to
enable UTF-8 encoded names. I guess the MD5 will change. This was never
officially supported but I know that if the MD5 changes here this will
harm them.

I wonder it's time to update the --migrate-md5-using-nametrans to a more
generic version. Or introduce yet another CLI option. I don't know.

>          # Cache the full folder path, as we use getfullname() very often.
>          self._fullname = os.path.join(self.getroot(), self.getname())

-- 
Nicolas Sebrecht