Validating all fields of a row & send errors to ErrFile/

[ray.wurlod]

The usual technique is to capture them individually in stage variables then use other stage variables to assemble the output message.

[WORKING_GRUNT]

Hi, Thanks for replying so fast.

I am capturing each field individually and storing them in separate stage variables but I cannot seem to get more then 1 error message per row. I am rather new to DS and there may be something subtle that I not aware of with this technique.

For example ...
Acct# Chk# ID AcctCde ChkAmt
123A 2222 ABC 100.00

If there is an error in the account number field and the account code field I would like to be able to format the following records

ErrDesc Acct# Chk# ID AcctCde ChkAmt
Non-Numeric. Acct# 123A .....................................
AcctCde Required 123A ....................................

I only get the last message, I cannot get the first error message.

Again any help would be most appreciated. Sorry for the long winded post.

Thanks ... Working Grunt. :D

[rleishman]

I have seen an idea posted in this forum whereby you concatenate the messages with a line feed character (ASCII 10) between each one, and them write them directly to a sequential file.
So, if a single input row has 4 errors, a single "row" is written to the output but it actually comes out as 4 rows.
A file is the ONLY place you can write them. If you direct the output to any other kind of stage it won't work.

[jenkinsrob]

If you have 3 errors on the same row in your source file do you want three rows in your Oracle table or 1 row detailing the 3 source errors??

[ray.wurlod]

Create validation stage variables of the kind

Code: Select all

If test Then " " Else "Column x is bad because..."

Note that each has a space character if the column is good.

To create a single message with all, simply concatenate these all together, then apply a Trim() function to reduce any extraneous internal space.

Code: Select all

Trim(svValidate1 : svValidate2 : svValidate3)

[vmcburney]

And you can put a char(10) at the start of each test message to make it appear on a seperate row. You can output both an error header file and an error item file, the header has some columns that identify the job name and process information (process identification and date) and something that identifies the row being processed. The item rows have details about the errors that occurred on that row. The char(10) format the output so the items follow the headers to make the file readable or loadable into a database.

[kduke]

I posted a routine that does this called KgdDebugMetadata. I think it is also on my tips page. The sequential file needs to be small because it reads it into a variable. It could be changed to readseq if you want. If you can figure it out then use it.

[WORKING_GRUNT]

To all who replied ....

[color=blue][size=9]Thank you![/size] [/color]

I am in the process of redesigning my job to accommodate what was suggested.

Thanks again for all your help.

:D