DSXchange

ray.wurlod wrote:or a kind of cheese originating in the Netherlands.

Edaaam..... Hmmmmmmmm....

Wow! Thanks for passing this on. One of the other DS developers just asked me this afternoon if I had come across something like this. I read this post, it was identical to her problem. Folks, if you solve problems to your own posts, or come across something like Andy did, PLEASE POST IT HERE! THANK...

What if I was to get the extract development team to use ascii(xFE) for field delimiters and ascii(xFF) for EOR markers?

This would bump into the values for @FM and @IM, but if I stay away from hash files, would I be OK?

Just kicking around other ideas....

chulett wrote:Can someone throw me a bone?

Me too. I was thinking Enterprise Data Mart but that seems a contradiction in terms.

Rob

SELECT TO_CHAR(UDT_DATE, 'YYYY-MM-DD HH24:MI:SS'), TO_CHAR(DATETIME_ADDED, 'YYYY-MM-DD HH24:MI:SS'), to_char(SYSDATE, 'YYYY-MM-DD HH24:MI:SS') AS LAST_UDT FROM edw_product_d Rather than do date manipulation in DS, (which I avoid at all costs!), do it in your SQL, i.e. SELECT TO_CHAR(UDT_DATE, 'YYYY...

Hmmm.....

Some interesting ideas here Kenneth.

Thanks.

Rob W

Ray,

We have used CFF in previous projects.

My problem is
1. doing some pre-processing in terms of validation and
2. telling the CFF how to tell where one field ends and an other starts, as well as where one record ends and the next one starts.

Rob

OK, thanks for the additional info Craig.

Rob

If you have the number of columns that are actually present in the file then you could read the file in a single row. Use the field function, or ereplace function to deal with that martian delimiter and after you find n number of delimiters (where n will be your column count), throw in a carrige re...

chulett wrote:
Are you on a Windows server like your first post indicates?

Yup.

Yessss, Craig, I have seen that, buuuttttt..... How do I determine which carriage return is actually the correct one? Thinking through this a bit more, I guess it would have to follow my last delimiter, i.e. I would have to specify that the last column is followed by the delimiter and the 'real' car...

OK, Folks, I've spent about an hour and a half browsing all the posts and have not found anything that gives me a solution to the particular task I have been set. Next week I will be receiving a peculiar file that will contain several fields that can be up to 32k wide. These fields can hold almost a...

Rganda, I have just completed working on a project that does exactly what you are describing. We used the following approach: - Read in the file using a single column (2000 wide in our case -> max record width plus 25% or so for future growth) to do preliminary validation such as header check, dupli...

narasimha,

I'd be very interested in your eventual solution. I will be receiving some files that have multiple 32k wide columns separated by a 3 character delimiters.

Please post your findings!

Assuming that the source is a flat file, run the data through an aggregator, grouping by all the fields except the date field then select the max date. Only one row will survive the aggregation.

If the source is an RDBMS, you could do something similar using a group by in the SQL.

HTH,

Rob W