Case of letters that specify encodings in MIME-encoded words

Case of letters that specify encodings in MIME-encoded words

[TSUCHIYA Masatoshi]

[-- Attachment #1: Type: text/plain, Size: 1785 bytes --]

Hi,

I noticed that several MUAs (including Microsoft Outlook Express) fail
to decode MIME-encoded subjects which are composed with Gnus.  In
order to check the detail of this problem, I send three mails that
have MIME-encoded subjects to my friend who uses Microsoft Outlook
Express, and ask him to check whether these subjects are successfully
decoded or not.  The result is here:

        MIME-encoded subject field      |   Result
  --------------------------------------+-------------
    =?iso-2022-jp?b?GyRCNEE7ehsoQg==?=  |  undecoded
    =?iso-2022-jp?B?GyRCNEE7ehsoQg==?=  |   decoded
    =?ISO-2022-JP?B?GyRCNEE7ehsoQg==?=  |   decoded

This result shows that this broken decoder decodes MIME-encoded words
whose encodings are specified in the uppercase letter "B", and shows
that it does not decode a word whose encoding is specified in the
lowercase letter "b".

I think that a careless programmer implemented this decoder reading
the fourth section of RFC2047,

   4. Encodings
      Initially, the legal values for "encoding" are "Q" and "B".

and he/she failed to notice the second section.

   2. Syntax of encoded-words

        encoded-word = "=?" charset "?" encoding "?" encoded-text "?="
        charset = token    ; see section 3
        encoding = token   ; see section 4

      Both 'encoding' and 'charset' names are case-independent.  Thus
      the charset name "ISO-8859-1" is equivalent to "iso-8859-1", and
      the encoding named "Q" may be spelled either "Q" or "q".

It is true that my reporting problem is not caused by Gnus, but I
think that the attached change may be useful for Gnus users who are
forced to communicate with users of such broken MUAs, and that it is
not harmful to Gnus at least.

Can I commit this change?

--
TSUCHIYA Masatoshi

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: Type: text/x-patch, Size: 389 bytes --]

--- rfc2047.el	7 Apr 2004 19:41:45 -0000	7.11
+++ rfc2047.el	7 May 2004 09:07:53 -0000
@@ -406,7 +406,7 @@
 			   'Q))))
 	 (start (concat
 		 "=?" (downcase (symbol-name mime-charset)) "?"
-		 (downcase (symbol-name encoding)) "?"))
+		 (upcase (symbol-name encoding)) "?"))
 	 (factor (case mime-charset
 		   ((iso-8859-5 iso-8859-7 iso-8859-8 koi8-r) 1)
 		   ((big5 gb2312 euc-kr) 2)

Re: Case of letters that specify encodings in MIME-encoded words

[Katsumi Yamaoka]

>>>>> In <87fza84xka.fsf@pine.kuee.kyoto-u.ac.jp>
>>>>>	TSUCHIYA Masatoshi wrote:

> Hi,

> I noticed that several MUAs (including Microsoft Outlook Express) fail
> to decode MIME-encoded subjects which are composed with Gnus.  In
> order to check the detail of this problem, I send three mails that
> have MIME-encoded subjects to my friend who uses Microsoft Outlook
> Express, and ask him to check whether these subjects are successfully
> decoded or not.  The result is here:

>         MIME-encoded subject field      |   Result
>   --------------------------------------+-------------
>     =?iso-2022-jp?b?GyRCNEE7ehsoQg==?=  |  undecoded
>     =?iso-2022-jp?B?GyRCNEE7ehsoQg==?=  |   decoded
>     =?ISO-2022-JP?B?GyRCNEE7ehsoQg==?=  |   decoded

Isn't there the same problem in the Latin world?  It's easy to
suspect that "=?iso-8859-1?q?..." may not be decoded either,
isn't it?  I sometimes see those who don't want Gnus to encode
Latin characters in headers.

> This result shows that this broken decoder decodes MIME-encoded words
> whose encodings are specified in the uppercase letter "B", and shows
> that it does not decode a word whose encoding is specified in the
> lowercase letter "b".

> I think that a careless programmer implemented this decoder reading
> the fourth section of RFC2047,

>    4. Encodings
>       Initially, the legal values for "encoding" are "Q" and "B".

> and he/she failed to notice the second section.

>    2. Syntax of encoded-words

>         encoded-word = "=?" charset "?" encoding "?" encoded-text "?="
>         charset = token    ; see section 3
>         encoding = token   ; see section 4

>       Both 'encoding' and 'charset' names are case-independent.  Thus
>       the charset name "ISO-8859-1" is equivalent to "iso-8859-1", and
>       the encoding named "Q" may be spelled either "Q" or "q".

> It is true that my reporting problem is not caused by Gnus, but I
> think that the attached change may be useful for Gnus users who are
> forced to communicate with users of such broken MUAs, and that it is
> not harmful to Gnus at least.

> Can I commit this change?

> --- rfc2047.el	7 Apr 2004 19:41:45 -0000	7.11
> +++ rfc2047.el	7 May 2004 09:07:53 -0000

Although it is against my will, I think it is needed.
-- 
Katsumi Yamaoka <yamaoka@jpl.org>

Re: Case of letters that specify encodings in MIME-encoded words

[TSUCHIYA Masatoshi]

>> On Mon, 10 May 2004 15:15:49 +0900
>> tsuchiya@namazu.org (TSUCHIYA Masatoshi) said as follows:

> Can I commit this change?

Because no objection is posted to this group in two days, I assume
that my proposal is approved tacitly.  So, it has been committed.

-- 
TSUCHIYA Masatoshi