Shell Programming and Scripting

I currently use a unix script to generate a pdf file by calling rwrun.sh. THe data is taken from database with nls_lang = utf8. Currently my UNIX script uses GERMAN_GERMANY.WE8ISO8859P1 and I am getting inverted question marks when I have other language text in my report. What kind of nls_lsng set up should I have so that I can get rid of those inverted question marks and print proper text?

Have you tried setting NLS_LANG to GERMAN_GERMANY.WE8ISO8859P1 before you run the database query to select the data out of the database? That should cause the proper characterset conversion to occur. If you still get the inverted question marks, then the data was corrupt in the database already, or the WE8ISO8859P1 characterset doesn't include the characters.

Thanks for the reply !

The NLS_LANG before getting data in UNIX is GERMAN_GERMANY.WE8ISO8859P1. The data is nto corrupted in database. It is a proper one.I could access the data promptly using a oracle report.

Yes the GERMAN_GERMANY.WE8ISO8859P1 does not support UTF format. I even tried GERMAN_GERMANY.UTF8. But it still didnt work.

in windows I have a file uifont.ali where I gave the instructions as

[ pdf: subset ]
"Arial Unicode MS" = "ARIALUNI.TTF"

Is there any uifont.ali file in oracle server in unix platform where I could implement this change?

What o/s was the client the data was inserted from into the database?
What is the code page of that client?
What is the NLS_LANG setting for that client?
Was the data created on that client or transferred to that client from another server?
If so, what is the characterset of the server the data was created on?

Inappropriate NLS_LANG settings can cause data corruption. However, the problem can be masked if you are selecting the data back out of the database from the client with the incorrect NLS_LANG setting. On that client, the data will appear perfectly normal, but only because the same incorrect characterset conversion is taking place coming out of the database that took place when the data was inserted into the database.

What you've said about the WE8ISO8859P1 characterset getting invalid results leads me to believe this is what has happened. If you can answer the questions above and bear with me while we do some troubleshooting, I believe we can figure it out.

It would be helpful if you could describe the process used to generate and insert data into your database as well.

1. Data is coming from the front end done in Java. It has UTF 8 set up

2. Yes the data is correctly transferred to the database. This I conirmed by selecting data from database using PL/SQL developer. So data is not corrupted in the DB

3. More over when I access the same data using oracle reports, they get displayed properly. All the above mentioned actions are done in windows

4. When I try to generate a PDF report from a unix shell script using the multilanguage data, I can correctly get one language and another one becomes inverted question marks. So this problem occurs in UNIX only. I even tried to set upthe NLS_LANG = AMERIACAN_AMERICA.UTF-8/ UTF8. Nothing works.

I believe I have almost answered all ur questions

Try setting the NLS_LANG on your unix box to same as it is on the windows server. Once the file is created, change your NLS_LANG back to what it was and transfer the file and open it on your windows server and see if it looks correct.