reading a fixed-width flat file

kvnbabu · Post by **kvnbabu** » Thu Jan 03, 2008 5:28 am

[color=blue]
Hi,
I have fixed width flat file with only one column of size 618 char. I have no problems in reading the correct record, but there can be invalid records with either less or more than 618 char length. I want to identify these bad records and route them to a different file. Could you suggest me how I do this?

I defined a column with 1000 char as size, defined the delimiter '000' (which is equiavalent to no delimiter), but somehow its not working out.
[/color]

ArndW · Post by **ArndW** » Thu Jan 03, 2008 5:40 am

Unless your fixed-width file has record delimiters (i.e. <cr><lf>) you cannot read it if some records have different lengths. So, does your file have line/record delimiters?

devidotcom · Post by **devidotcom** » Thu Jan 03, 2008 5:44 am

You could have a reject link from the file from which you are trying to read. The invalid records of variable length will sent to the reject link file.

kvnbabu · Post by **kvnbabu** » Thu Jan 03, 2008 5:45 am

[quote="ArndW"]Unless your fixed-width file has record delimiters (i.e. <cr><lf>) you cannot read it if some records have different lengths. So, does your file have line/record delimiters?[/quote]

New line character can be considered as end-of-record.

kvnbabu · Post by **kvnbabu** » Thu Jan 03, 2008 6:02 am

problem is not with rejecting them. but i am missing something, while reading the source files, the new line character also comes along the data and I am unable to strip it off.

ArndW · Post by **ArndW** » Thu Jan 03, 2008 6:16 am

Since you have a terminator, what devidotcom suggested will work perfectly.

For the <lf> error - Have you declared the <cr><lf> in your sequential file stage definitions for the "record delimiter"?

kvnbabu · Post by **kvnbabu** » Thu Jan 03, 2008 6:27 am

devidotcom,

Please could you eloborate on what is to be done?

Thanks

ray.wurlod · Post by **ray.wurlod** » Thu Jan 03, 2008 2:34 pm

Look on the Stage properties of your Sequential File stage - is it set for UNIX-style, DOS-style or None? Look in the file itself, using a hex editor. How are lines in the file actually terminated?

Note that, if lines can be more, equal or less than 637 characters, then you do NOT have fixed-width format. Therefore you need to read using a single VarChar field, and test in a downstream Transformer stage whether the length is correct (and anything else you might want to test).

Server jobs do not support reject link on Sequential File stage; the advice from devidotcom relates only to parallel jobs.