SonarQube 6.3 LDAP/SSO UTF-8 encoding - encoding

Question

We‘re using LDAP/SSO in my company which provides the username in UTF-8 format to SonarQube.
However LDAP/SSO sends the username in UFT-8 but SonarQube requires Latin1/ISO-8859. There is no way to change the encoding on LDAP/SSO or SonarQube.
Result wrong umlauts:
Andrü Tingö = Andr«Ã Ting¼Ã OR äëüö = äëüÃ
Is there any workaround?

Answer 1

I wanted to post this as comment, but I need 50 reputations to write comments.
We are using simplesamlphp for SSO as IdP and SP. IdP takes cn, givenName and sn from LDAP, which has UTF-8 values. Loginnames/Usernames are us-ascii only.
If the user comes to Sonar, the non-us-ascii characters are incorrect - they were converted from ... to utf-8, even they already are in utf-8.
If I use the attributes from IDP in PHP which sends the page in UTF-8, then characters are correct.
I did just now one test. In our Apache Config we set the X-Forwarded-Name to MCAC_ATTR_CN attribute what SP get from IdP. Original configuration is:
RequestHeader set X-Forwarded-Name "expr=%{reqenv:MCAC_ATTR_CN}"
Now I have added fixed string in UTF-8:
RequestHeader set X-Forwarded-Name "expr=%{reqenv:MCAC_ATTR_CN} cäëöüc"
The "c" characters are only separators to see the encoded text better.
The hexdump of this configuration line is:
0000750: 09 0909 5265 7175 6573 7448 6561 ...RequestHea
0000760: 6465 7220 7365 7420 582d 466f 7277 6172 der set X-Forwar
0000770: 6465 642d 4e61 6d65 2022 6578 7072 3d25 ded-Name "expr=%
0000780: 7b72 6571 656e 763a 4d43 4143 5f41 5454 {reqenv:MCAC_ATT
0000790: 525f 434e 7d20 63c3 a4c3 abc3 b6c3 bc63 R_CN} c........c
00007a0: 220a ".
As you can see, there are fixed utf-8 characters "ä" c3a4 "ë" c3ab "ö" c3b6 "ü" c3bc.
From LDAP comes follwing name:
xxxxxx xxxxx xxxx äëüö
In Apache config is appended " cäëöüc", therefore resulting name should be:
xxxxxx xxxxx xxxx äëüö cäëöüc
But in Sonar, the name is displayed as
xxxxxx xxxxx xxxx äëüö cäëöüc
You get similar result if you convert follwing text:
xxxxxx xxxxx xxxx äëüö cäëöüc
from ISO-8859-1 to UTF-8:
echo "xxxxxx xxxxx xxxx äëüö cäëöüc" | iconv -f iso-8859-2 -t utf-8
xxxxxx xxxxx xxxx äÍßÜ cäÍÜßc
The "¤" character is utf-8 char c2 a4:
00000000: c2a4 0a ...
I have made tcpdump on loopback to get communications from apache proxy module to sonarqube and even there you can see correct UTF-8 characters c3a4 c3ab c3bc c3b6 comming from IdP and then between "c"s you can see c3a4 c3ab c3b6 c3bc comming direct from apache.
00000000 47 45 54 20 2f 61 63 63 6f 75 6e 74 20 48 54 54 GET /acc ount HTT
...
00000390 58 2d 46 6f 72 77 61 72 64 65 64 2d 4e 61 6d 65 X-Forwar ded-Name
000003A0 3a 20 72 6f 62 65 72 74 20 74 65 73 74 32 20 77 : xxxxxx xxxxx x
000003B0 6f 6c 66 20 c3 a4 c3 ab c3 bc c3 b6 20 63 c3 a4 xxx .... .... c..
000003C0 c3 ab c3 b6 c3 bc 63 0d 0a ......c. .
...
The system has locales set to en_US.UTF-8, if this matters.
So Sonar gets really UTF-8 Text from Apache (direct config or from IdP) but then something probably converts this utf-8 text as if it was iso-8859 text to utf-8 again and makes nonsense.
Do you have any idea now? Could this be something in sonar or in wrapper or somewhere some options set incorrectly?
Regards,
Robert.