Chinese emails

[Paul]

Antony Stone wrote:
> On Thursday 12 November 2009 09:59, Paul wrote:
>
>   
>> Hi
>>
>> Ive been down this road before and thought I fixed it, however am
>> getting some complaints from users again that chinese email is being
>> blocked as the filetype rules are saying its a program.
>>
>> When I run just "file" on the message it returns its an executable,
>> however "file -i" returns its text/plain
>>
>> [root at geneva nAB8qd2B008626]# file msg-23153-1986.txt
>> msg-23153-1986.txt: COM executable for DOS
>> [root at geneva nAB8qd2B008626]# file -i msg-23153-1986.txt
>> msg-23153-1986.txt: text/plain; charset=iso-8859-1
>>     
>
> Both of those would seem to be incorrect to me, since you say it isn't an 
> executable, but it can't possibly be ISO-8859-1 if it's in Chinese...
>
>   
>> I have added
>>
>> allow   -   text/plain   -   -
>>
>> To the top of my filetype.rules.conf - however the message is still
>> being blocked with "No Programs Allowed"
>>     
>
> Have you changed "File Command" in MailScanner.conf to include the -i option?
>
>   
>> Im running version 4.74.16.
>>
>> Any ideas?
>>     
>
> "man file" suggests that one of the things the -i option does is tell file to 
> use a different source of magic data (/usr/share/file/magic.mime instead 
> of /usr/share/file/magic on my Debian system) - perhaps you could just rename 
> these files to achieve what you want without needing the -i option?  The 
> files look to be compatible in format with each other.
>
>
> Regards,
>
>
> Antony.
>   

When I release the email it certainly looks like Chinese text - there is 
no executable attachment.

This has been brought up before on the list, I don't think the issue was 
limited to Chinese characters either.

My understanding was you did not need to add the -i option in 
MailScanner.conf, this is an extract from filetype.rules.conf,

# An optional fifth field can also be added before the "log text", which
# makes the checked text check against the MIME type of the attachment
# as determined by the output of the "file -i" command.

I thought if the fifth field is present it will be checked with -i?

I'm certain I fixed this some time ago and I cant think of anything I 
have done to break it!

Kind Regards

Paul