Question

I have a multi-column table of data, each row unique, and I want to know how to sort it based on multiple columns. If sorting alphabetically a solution is already described here. However, instead of alphabetical sorting, I need to sort each column based on a custom ordering stored in another list. For example, if my table is

mytable = [
    ('A1', 'B1', 'C1'),
    ('A1', 'B2', 'C2'),
    ('A2', 'B2', 'C1'),
    ('A2', 'B2', 'C2')
]

I might want the first column to be ordered [‘A2′,’A1’], the second column to be ordered [‘B1′,’B2’], and the third column to be ordered [‘C2′,’C1’]. The proper result would be

mytable = [
    ('A2', 'B2', 'C2'),
    ('A2', 'B2', 'C1'),
    ('A1', 'B1', 'C1'),
    ('A1', 'B2', 'C2')
]

Answer 1

This will do what you’re looking for:

orderings = (
    ('A2', 'A1'),
    ('B1', 'B2'),
    ('C2', 'C1')
)

orders = [dict([(v, i) for i, v in enumerate(o)]) for o in orderings]

mytable.sort(key=lambda r: tuple(o[c] for c, o in zip(r, orders)))

In practice, the columns might not all be sorted, or might be sorted in a priority other than left-right order. That could be solved by attaching an index to each and adapting the algorithm accordingly.