directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Emmanuel Lecharny (JIRA)" <j...@apache.org>
Subject [jira] Commented: (DIRSERVER-873) Apparent problem in decoding LDAP requests
Date Fri, 16 Mar 2007 13:45:09 GMT

    [ https://issues.apache.org/jira/browse/DIRSERVER-873?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12481614
] 

Emmanuel Lecharny commented on DIRSERVER-873:
---------------------------------------------

Ok, I went through some more investigation, and here is what I get :
- DN are stored in two form inside the server : bytes and user provided form (Unicode). The
bytes are always 7 bits, so the 'é' is stored as six bytes : 0x5C, 0x43, 0x33, 0x5C, 0x41,
0x39 ("\C3\A9")
- when we return a response (SearchResultEntry), those six bytes are translated to two bytes
: 0xC3 0xA9, which are the UTF-8 bytes for 'é'. 

So basically, I would say that the server behavior is correct. I have tested it on my mac
osx, with LdapStudio, and the only bias I can see is that my mac is using french code (ISO
8859-1).

more to come.

> Apparent problem in decoding LDAP requests
> ------------------------------------------
>
>                 Key: DIRSERVER-873
>                 URL: https://issues.apache.org/jira/browse/DIRSERVER-873
>             Project: Directory ApacheDS
>          Issue Type: Bug
>          Components: asn1, ldap
>    Affects Versions: 1.5.0
>         Environment: WindowsXP, JDK 1.5.06
>            Reporter: Norval Hope
>         Attachments: Copy of apacheds-rolling.log
>
>
> I'm sending a request with a non-seven bit ascii char in a DN from JXplorer and it seems
to be decoded incorrectly when it arrives at ApacheDS (ASN level byte logging is attached).
I see the same problem when sending the request from JMeter too.
> The DN I used was "uid=tté,ou=system" (same as uid= value in attributes) but decoded
as 'uid=tt\C3\A9,ou=system' where I was expecting 'uid=tt\E9,ou=system'. The full data I sent
was:
> DN: uid=tté,ou=system
> cn: tté
> objectClass: inetOrgPerson
> sn: ttsn
> I tried sending the DN as 'uid=tt\E9,ou=system' but then JXplorer tried to quote the
'\' itself so it seems that it wants to be responsible for the conversion to an rfc2253 compliant
DN itself.
> I know all this encoding/codepage/locale/lang stuff is always a big PITA and that this
may be related to to the default encoding on WinXP or some such thing, but even then I'm struggling
to see how one Latin char 'é' becomes two hex chars \C3 and \A9 in UTF8. At any rate I have
attached the ASN byte logs so that someone more knowledgeable can review and state definitively
that the bytes received by AD are already "bad". Note that I ran this test against the latest
1.5 trunks (after first noticing the issue in my own out-of-date a heavily modified version
of 1.5)

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message