So whats the right approach when we have records in the header as opposed to not having them in details?
If I read your question right, you're asking how to address the headers in a file? I haven't run across one of these multi-format files that had a header, because a single header can't describe the various metadata. Perhaps a file could have headers within
the data - I haven't seen this approach, but if this exists you'd need to use your script to detect the header records (the method for which would vary depending on the makeup of the header) and either discard the headers or use them to identify the upcoming metadata.
Tim Mitchell, Microsoft Data Platform MVP
Data Warehouse and ETL ConsultantTimMitchell.net
| Tyleris.comETL Best Practices