Originally posted by: intp

 Hi,

We have a map (in WTX 8.4) that reads a database with different columns (Unicode data) and we've found a problem with a character.

The input type tree has UNICODE Little Endian, delimiter '|' and release character '!'. The delimiter and release character has native language, the way they are generated with the importer.

The row 48 is correctly read, but the row 49 has a problem with the third column, because there's a character interpreted as a release character and then the rest of columns are read incorrectly.

If we don't put any release character, the row is correctly read. But we don't know if a field can store data that needs a release character when is processed in the mapping.

Can anyone help on it?

Thanks in advance.

Originally posted by: paul.brett

The Database Interface Designer imports Typetrees, and it is up to the end user to manually edit them to suit the database.

In this case, you would edit the properties of the field objects, as well as the release characters from Native to UTF-16.

I have tested this scenario, and it works well, both for your example data, and also for data that contains a character that would need to be escaped ('ЦПС!ЦПС').

Thank-you.

Paul.

Follow me on Twitter

Originally posted by: paul.brett

Seems perhaps I was a little hasty in my previous statements.  Further testing shows that inclusion of a character that is supposed to be released is not working, with any subsequent data in this field being truncated.

I suggest you raise a PMR for this issue, as I suspect it might be a bug.

Thank-you.

Paul.

Follow me on Twitter

Originally posted by: intp

ok, thanks Paul.

Originally posted by: paul.brett

OK, I have got it working now.  It seems with an NLS_LANG value of AMERICAN_AMERICA.AL32UTF8 and the Typetree text field objects and the release character set to UTF-8, I can insert and retrieve this data ('ЦПС!ЦПС') successfully.

I have included the 'backup' files (Out1.bak from the database insert Test1 and In1.bak from the database query Test2) so that the 'raw' data can be seen, with the double bang (!) characters.

Thank-you.

Paul.

Follow me on Twitter

Attachment(s)

Unicode.zip   8 KB 1 version

Originally posted by: intp

But I can't change the NLS_LANG due to other existing mappings... Is there any way that only affects this mapping or the way it calls the data?

Originally posted by: paul.brett

None that I know of.  Your workaround is going to work fine, right up until there is a character in the data that needs to be released, at which point, the input will either not validate, or (worse) your map will complete with garbage thrown out the other side.

Can this map be run in a separate Launcher, that has the environment set correctly for UTF-8 data?

Thank-you.

Paul.

Follow me on Twitter

Originally posted by: intp

Currently I can't use another Launcher in other server, with other language environment. 

I'm also openning a PMR.