[sqwebmail] Problems with corrupted japanese text.

[Nathan Ollerenshaw]

Hey there,

I configured the latest stable version of sqwebmail (3.6.0). Configure 
line below:

./configure --with-cachedir=/apps/webmail-test/sqwebmail/cache
  --with-cacheowner=vmail
  --enable-cgibindir=/apps/webmail-test/cgi-bin
  --enable-imagedir=/apps/webmail-test/htdocs/images
  --enable-imageurl=http://wm2.sh.test.intra:88/images
  --enable-mimetypes=/apps/webmail-test/apache/conf/mime.types
  --enable-unicode --prefix=/apps/webmail-test/sqwebmail
  --enable-mimecharset=UTF-8
  --enable-softtimeout=3600
  --enable-hardtimeout=7200
  --with-module=authldap

I applied no patches to the code. The japanese templates (a little 
outdated I think) were installed. They were converted to UTF-8 (using 
BBEdit) and the CHARSET file was updated to reflect this. (I did this 
because in iso-2022-jp, the templates would show as corrupted in some 
places, and converting to UTF-8 fixed this problem).

I can log in with English or Japanese fine, there is no corrupted text 
in the templates in Japanese mode.

These are the problems I'm experiencing.

I'm using the following text to test. It means "Japanese language test".

日本語　テスト　日本語　テスト　日本語　テスト　日本語　テスト　日本語　テスト

Copy this a few times on one line, and use it in the subject.

Hopefully your terminal/email client supports UTF-8 :)

1) If I send a message in Japanese (in either template) to myself, 
reading the message in sqwebmail will show random corruption in the 
text.

2) Downloading the message in a mail client will show that the entire 
message is corrupted? Its almost like the message is being marked as 
UTF-8, but it really isn't.

3) Sending a message from my regular mail client works fine, it is 
encoded as iso-2022-jp and sqwebmail views it perfectly - which is not 
the problem - the problem happens when you forward that message, the 
subject line becomes corrupted and if there is no line breaks in the 
body, the body will become corrupted in a random fashion like above.

4) The address book will occasionally become corrupted when you create 
a new nickname, depending on the text you use. For example, using the 
japanese templates, I copy and paste the text that says "atarashii 
nickname" in japanese ("new nickname"). Normally, the screen would 
change with the new nickname displayed, and two boxes "name" and 
"address". However, the place where the nickname is displayed, loses a 
character and gets /font> appended. It's a bit hard to explain, but if 
you try copying various bits of Japanese text into the new nickname 
field, you should be able to replicate the problem. It's almost as if 
the last byte is being dropped from the unicode text.

Is anyone else having problems like this? Have any of you found a 
solution to these problems? Does anyone have any suggestions they can 
make towards fixing them? I'm willing to try any experimental patches 
necessary to try and fix it.

Any help at all would be appreciated.

Thanks for reading this long mail,

Regards,

Nathan.

-- 
Nathan Ollerenshaw - Systems Engineer - Shared Hosting
ValueCommerce Japan - http://www.valuecommerce.ne.jp

"The person who stands up and says, 'This is stupid,' either is
asked to 'behave' or, worse, is greeted with a cheerful 'Yes,
we know! Isn't it terrific!'." - Frank Zappa