Well, what DO you want to do with extra columns? Ignore them, process them specially? This is really a business question, I think. What do your customers (even if just another internal department or person) expect to happen?

The idea is this (I've used it many times):

1. The Flat File connection just outputs one column. Call it 'Data'.

2. The CS transform or Script transform (or both) examines the one column and decides what to do: Accept it and send it into the dataflow or reject it and send it to the error output. Note that the script component could set a variable or add a RecordType column to the data flow that a CS transform could use for further discrimination. That's one reason you may use both transforms in series.

3. If accepted, add Derived Column transformations to parse the one column 'Data' into the other columns you need. Note that, if you use a CS transform, you can have one path for "normal" inputs and one for inputs with extra columns.

4. Further process the data flow(s) and finally output the data somewhere (e.g. OLEDB destination)