How can I sort tuples by reverse, yet breaking ties non-reverse? (Python)

Question

If I have a list of tuples:

results = [('10', 'Mary'), ('9', 'John'), ('10', 'George'), ('9', 'Frank'), ('9', 'Adam')]

How can I sort the list as you might see in a scoreboard - such that it will sort the score from biggest to smallest, but break ties alphabetically by name?

So after the sort, the list should look like:

results = [('10', 'George'), ('10', 'Mary'), ('9', 'Adam'), ('9', 'Frank'), ('9', 'John')]

At the moment all I can do is results.sort(reverse=True), but breaks ties reverse alphabetically too...

Any help would be much appreciated. Thanks!

Answer 1

The simplest way to achieve what you want is to use the fact that python sort is stable. This allows to first sort alphabetically and then by score:

In [11]: results = [(10, 'Mary'), (9, 'John'), (10, 'George'), (9, 'Frank'), (9, 'Adam')]

In [12]: results.sort(key=lambda x: x[1])

In [13]: results.sort(key=lambda x: x[0], reverse=True)

In [14]: results
Out[14]: [(10, 'George'), (10, 'Mary'), (9, 'Adam'), (9, 'Frank'), (9, 'John')]

The first sort sorts alphabetically, in ascending order. The second sort sorts by score, in descending order, maintaining the relative order of elements with equal score.

You can do this to do even more complex sorts. Just remember that you must first sort by the secondary key, and then by the first key. (If you have three keys, first sort by the third, then by the second, and lastly by the main key).

If you don't want to call sort twice you'll have to write a more complex key function. Something like:

In [50]: def key(elem):
    ...:     return elem[0], [-ord(c) for c in elem[1]]

In [51]: sorted(results, key=key, reverse=True)
Out[51]: [(10, 'George'), (10, 'Mary'), (9, 'Adam'), (9, 'Frank'), (9, 'John')]

In particular, every time you have something sorted in lexicographic order(such as strings, tuples, lists etc.), you can invert the order by changing the sign to all the elements.

Answer 2

sort method accept optional key parameter.

key specifies a function of one argument that is used to extract a comparison key from each list element

You need to convert string to number:

>>> results = [('10', 'Mary'), ('9', 'John'), ('10', 'George'), ('9', 'Frank'), ('9', 'Adam')]
>>> results.sort(key=lambda x: (int(x[0]), x[1]), reverse=True)
>>> results
[('10', 'Mary'), ('10', 'George'), ('9', 'John'), ('9', 'Frank'), ('9', 'Adam')]

Answer 3

You can have very granular control on how to sort each of the values in a tuple with sort method's key parameter. For example:

In [41]: results = [('10', 'Mary'), ('9', 'John'), ('10', 'George'), ('9', 'Frank'), ('9', 'Adam')]

In [42]: results.sort(key=lambda (score, name): (-int(score), name))

In [43]: results
Out[43]: 
[('10', 'George'),
 ('10', 'Mary'),
 ('9', 'Adam'),
 ('9', 'Frank'),
 ('9', 'John')]