Axel Luttgens wrote:

Le 8 juil. 09 à 12:04, Nikita Koshikov a écrit :

I don't know whether dovecot-sieve is supposed to do the conversion automatically, but it is sure that an RFC822 header MUST be in US-ASCII (so as to be 7bit-clean).

The required encodings are described in: http://www.rfc-editor.org/rfc/rfc2047.txt For example, a subject header with the french word "Résistance" could be encoded as: Subject: =?iso-8859-1?Q?R=E9sistance?=

Could you try by inserting the encoded subject line directly in the config file and see what gives? On the other hand, perhaps does sieve provide the needed tools itself; someone?

Indeed, this is currently not possible. For the new Sieve implementation this is on my TODO list, but the old CMUSieve that Nikita is using will probably never provide support for this.

Regards,

Stephan

On Wed, 8 Jul 2009 12:41:38 +0200 Axel Luttgens <AxelLuttgens@swing.be> wrote:

Le 8 juil. 09 à 12:04, Nikita Koshikov a écrit :

...

Hello list,

I have noticed that sieve vacation action breaks some non-english
letters in the subject. I tested with russian language, if subject contains letters from
specific diapason russian alphabet, the message received with
unreadable subject, but text in the body with the same letters looks
ok. Here is example of full subject header, it contains russian alphabet: Subject: абвгде�?жзийклмноп�?�?�?�?�?�? �?�?�?�?�?�?�?�?�?�?

Maybe this headers will be usefull: MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit

.dovecot.sieve file looks correct, here is it: require ["vacation"];

rule:[test]

if anyof (true) { vacation :days 1 :addresses ["user@domain.com"] :subject
"абвгдеёжзийклмнопрстуфхцчшщъыьэюя"
text: Тестовое русское сообщение(Test russian
message). ; }

I'm running dovecot-1.1.16 with dovecot-sieve-1.1.6 and
managesieve-0.10.6.

Please tell, is it possible to fix this ?

I don't know whether dovecot-sieve is supposed to do the conversion
automatically, but it is sure that an RFC822 header MUST be in US- ASCII (so as to be 7bit-clean).

The required encodings are described in: http://www.rfc-editor.org/rfc/rfc2047.txt For example, a subject header with the french word "Résistance" could
be encoded as: Subject: =?iso-8859-1?Q?R=E9sistance?=

Could you try by inserting the encoded subject line directly in the
config file and see what gives?

This made a trick, I have tried variant with =?UTF-8?B? - that worked for me, but this is not "user way". I think, the best way will be to exlain my users not to put non-english characters in the subject.

On the other hand, perhaps does sieve provide the needed tools itself;
someone?

Axel

Thanks everyone for replies.

On Sat, 11 Jul 2009 07:24:51 +0200 (CEST) "Benny Pedersen" <me@junc.org> wrote:

On Wed, July 8, 2009 12:04, Nikita Koshikov wrote:

...

Please tell, is it possible to fix this ?

fix is nedded in sieve client what ever that is, not in anything dovecot does

so to test: iconv the sieve script to unicode, and save it, then it works ?

Sieve script has already been in UTF encoding and this not working. The problem is to convert e-mail headers into the right format, as Alex wrote. If convertion is made "by hands" everything works as expected.

-- xpoint

On Fri, 2009-10-16 at 11:48 +0400, Proskurin Kirill wrote:

Is this problem still not fixed in new sieve implementation?

I just implemented message header encoder: http://hg.dovecot.org/dovecot-1.2/rev/bd4a6f500c75

So only thing left to do is for Stephan to start using it. :)

On Wed, 2009-10-28 at 12:32 +0100, Stephan Bosch wrote:

1: Test 'UTF-8 Subject' FAILED: subject header is not encoded/decoded properly: expected: Auto: Я могу есть стекло, оно мне не вредит. decoded: Auto: Я мог�? е�?�?�? �?�?екло, оно мне не в�?еди�?.

The bug was actually that str_sanitize() didn't understand UTF-8 input. Fixed.