Regarding the post that talked about mainframe file dumps that used column width as the delimiter vs csv, pipes, tabs etc. I run across this a lot because the folks that actually purchase our data absolutely refuse to even discuss data formats - it's a 'not my problem to load it issue'.

There's no easy answer, but the sweat equity answer is to build a parser in the programming language of you choice to replace two spaces with one (except when found between quote delimiters). Run that multiple times until there are no double spaces, then replace the space with the column delimiter of your choice.

Now if I could just get mainframe weenies to understand that giving me a data dictionary of the columns, data types and widths IN column name ALPHABETICAL ORDER vs. ordinal position doesn't help at all....(especially when the first row of the file is not column names, which for some reason just never happens on main frame files).