Question

I am trying to sort a recursively sort a 2 level nested default dict. I have not been able to figure out how to do this properly. My problem statement is given below:

1. First level key must be a natural sort.

2. Second level key must be sorted in a specific order. I have tried to create a list whose indices represent the order of the elements. My code snippet is given below:

   import operator
   import collections
   trade_group_totals = collections.defaultdict(lambda:collections.defaultdict(float))    
   trade_group_totals['foo']['ABC'] = 100
   trade_group_totals['foo']['XYZ'] = 50
   trade_group_totals['bar']['ABC'] = 150
   trade_group_totals['bar']['XYZ'] = 250
   

My sorting index comparison:

trade_groups = ['XYZ', 'ABC']

def TradeGroupSort(trade_group):
   return trade_groups.index(trade_group)   

def SortTotals(totals, sort_function_one, sort_function_two):
  return [
      (k1, v1) for k1, v1 in [(k, sorted(v.iteritems(), key=sort_function_two))
           for k, v in sorted(totals.iteritems(), key=sort_function_one)]]

I am invoking the function as follows:

SortTotals(
    trade_group_totals, operator.itemgetter(0),
    sort_function_two=lambda x: operator.methodcaller('TradeGroupSort', x))

My expected output should be:

[('bar', [('XYZ', 50), ('ABC', 100)]), ('foo', [('XYZ', 250), ('ABC', 150)])]

But the generated output is

[('bar', [('XYZ', 50), ('ABC', 150)]), ('foo', [('ABC', 150), ('XYZ', 250)])]

Answer 1

Unfortunately, none of the solutions worked for me. This is the UGLY sorting that I came up with.

def TradeGroupSort(trade_group):
  """Sorting the trade group."""
  group, _ = trade_group  # The value is a tuple ('XYZ', 50) for e.g.
  return trade_groups.index(group)

I could not think of a better solution than that. I invoked this function as follows:

SortTotals(trade_group_totals, operator.itemgetter(0), TradeGroupSort)

This returns the expected result.