The smallest order

There are n tasks, numbered 0 to n − 1. Each edge (u, v) says task u must come before task v. Return an order of all n tasks that keeps every rule (a topological order). If several orders are possible, return the smallest: the one with the smallest first task; among those, the smallest second task; and so on. This is how Python compares lists. If no order is possible, return None.

edges = [(5, 2), (5, 0), (4, 0), (4, 1), (2, 3), (3, 1)]
smallest_order(6, edges)       # [4, 5, 0, 2, 3, 1]
smallest_order(3, [(1, 0)])    # [1, 0, 2]
ring = [(0, 1), (1, 2), (2, 0)]
smallest_order(3, ring)        # None

Limits

• Up to 100,000 tasks and 200,000 rules. A rule may be given twice.
• Looking through all n tasks for the smallest ready one at every step costs about n² steps: far too slow. The big tests have 3 seconds each.