DSXchange

Hi,
Can we convert a Binary feed to ASCII feed in UNIX? Can you pls provide a sample code, if available?

We have a Datastage job to 'split' the Binary feed containing Header, detail recods and Trailer record. Thought is to replace the Datastage 'split' job to a UNIX script which can parse the feed and separate the Header, Detail and Trailer record.

Thanks.

Depends on what you mean by "binary feed". I'm guessing you mean a mainframe file either in EBCDIC or with packed fields or both. Can you clarify? And confirm you are using a Server job, please.

Not too sure what you are looking for but I think "iconv" can do the job for you.

Syntax: iconv -f encoding -t encoding

The job includes the need to "parse the feed and separate the Header, Detail and Trailer record", not just a straight conversion as noted in the subject...

My bad! Will learn to read the content before jumping to a solution next time

Ria,

My first question would be why is it necessary to accomplish the split in Unix? I don't see any performance gain to be had between (for example) dropping the header and trailer in Unix and dropping it after the input stage.

Nearly every job I have starts with an FTP stage from a mainframe dataset. When the dataset includes more than one record type, I parse the records in a transformer stage using substring derivations and the record type as a constraint. Complex Flat File does it on the input.

If your source is mainframe and EBCDIC, the FAQ on this may be helpful.

Thanks for your quick responses!

Yes, its EBCDIC file with packed data format containing heade, Detail and Trailer record. We've an existing DataStage Server job for splitting the file into different formats.

iconv has been used in UNIX. But But, its showing a message ""Can not open converter". Command:
iconv -f UTF-8 -t ASCII TPS_DDS_20120411.DAT
(UTF-8 was also checked with UTF-16, BIN also, but the error msg was the same). Can you pls share if above command needs to be corrected?

od command was also. But, it was not useful.
od -c TPS_DDS_20120411.DAT
od -x TPS_DDS_20120411.DAT

Thanks in advance for your immediate attention.

Stick with the DataStage job. Any straight 'conversion' of the entire file will corrupt the packed fields and it's more trouble than it is worth to try to parse out the different record types, converting the individual string fields and unpacking the others using a script. Of course, you could fall back on Perl or something of that ilk but if you've got something that works why bother?

I can't help with your command line questions. I can offer a successful method I use, which I summarized in the FAQ viewtopic.php?t=143596&sid=96688420e3bd ... f6045c5a9b

Read the beginning section, then look for the question "Q: My mainframe file has two or more record types. If I'm not using Complex Flat File, how do I read the file?"

Post any further questions you may have here, and I'll try to help.

Do keep in mind the fact that (as best as we can tell) they already have a DataStage job that does all this. They are looking for an alternative for some reason.

the quickest way would be to use datastage to do this task. However, there is option like dd to convert the ebcidic file to ascii, this does not convert compressed or comp fields. so, the best way is to write a c or cobol program that converts byte by byte.

alternatively, if you have perl and know how to use it then you could check this link which has got some code which does this taskhttp://search.cpan.org/~cxl/Convert-EBC ... /EBCDIC.pm

Hope this helps!!