Search Results

The NF inside awk counts the number of fields.
So, if your datafile has more columns, than that number would need to be adjusted.
Make sense?

OK,
it seems that your sed implementations doesn't interpolate the \t escape sequence.
I put it only to be safer, so you can just remove it:

sed 's/\([^0-9 ]\) *\([^0-9 ]\)/\1\2/g' infile