Most data from ATNF observatories distinguishes on and off source data using the file name. This makes it easy to create two scantables with the source and reference data. As long as there was exactly one reference observation for each on source observation for following method will work.
For Mopra and Parkes data:
ASAP> r = scans.get_scan('*_R')
ASAP> s = scans.get_scan('*_S')
For Tidbinbilla data
ASAP> r = scans.get_scan('*_[ew]')
ASAP> s = scans.get_scan('*_[^ew]')