Certain characters such as "-" in UTF-8 are not displayed correctly in file

[Former Member]

Hi,

Has anyone else seen a similar scenario?

There are certain characters entered in the R/3 system (UTF-8) such as a and ("), which looks different then the normal dash or quotation, in a free form text field. The values of these fields are transferred through XI via abap proxy and written to a mainframe file via the file adapter. The default file encoding of our XI system is ISO-8859-1. When the file is created these characters are interpreted as a (?). When we change the encoding to UTF-8 in the receiver adapter, it translates the characters as (b.). When I write the file out to the file system with encoding UTF-8, everything looks normal.

Can anyone shed some light on this situation?

Thanks.

Best Regards,

Duke

[Former Member · Accepted solution]

Hi,

This could be due to wrong code page. See Stefan's comment.

Regards,

Sarvesh

[stefan_grube]

Is this an XML file? Then you should mark the file type as "binary" and use the XMLAnonymizerBean for transferring the file to ISO-8859-1.

But I have no idea, why " and - are changed, as the code of these characters are the same in both codepages.

Is it correct in the SXMB_MONI?

Stefan

[Former Member]

Hi Stefan,

To clarify, the (") the user is entering are curly quotes. I found this on google.

Windows ANSI (224 characters; since Win3.1). This adds the accented characters common in Western European languages (including vowels with acute, grave, circumflex and umlaut accents), plus some other typographical symbols like u201Ccurly quotesu201D, bullets, currency symbols etc. The additional 126 characters above the ASCII set are called u201CUpper ANSIu201D.

This is a flat file using content conversion on the receiver file adapter. In SXMB_MONI it looks correct. If I do a view source, I can see that the quotations are the "curly quotes" and not the same as when I enter the character.

Best Regards,

Duke

[stefan_grube]

Hi Duke,

In that case the answer is easy: That sign simply does not exist in ISO-8859-1

See here:

http://en.wikipedia.org/wiki/ISO-8859-1

Chose encoding cp1252 (Windows 1252).

Didn't you mention Mainframe? You might have to choose the correct EBCDIC codepage.

In cases like this I always give the same advice: Figure out which codepage you really need on the receiver system, aks the administrator. Do not just guess.

Regards

Stefan

Edited by: Stefan Grube on Nov 25, 2008 8:51 PM

[Former Member]

Thanks Stefan for your help.