sort date objects in Python

Question

I start out with date strings:

from operator import itemgetter
import datetime as DT

# unsorted dates
raw = (map(int, "2010-08-01".split("-")),
       map(int, "2010-03-25".split("-")),
       map(int, "2010-07-01".split("-")))

transactions = []
for year, month, day in raw:
    new = (DT.date(year, month, day), "Some data here")
    transactions.append(new)
# transactions is now a list with tuples nested inside, for example
# [(date, "Some data here"), (date, "Some data here")]

sorted(transactions, key=itemgetter(0))

for info in transactions:
    print info

I get the following, and its not sorted according to date:

(datetime.date(2010, 8, 1), 'Some data here')
(datetime.date(2010, 3, 25), 'Some data here')
(datetime.date(2010, 7, 1), 'Some data here')

How do I sort these according to date?

Answer 1

sorted returns a sorted list, but the input itself remains unmodified. Try

transactions.sort(key=itemgetter(0))

instead. (By default tuples are compared lexicographically, so you don't need the key).

Answer 2

You should use sort method if you want to modify your original data, often it is better to preserve the original data and to sort by sorted() and assign the value in new variable.

transactions.sort()

The key itemgetter is unnecessary as you want normal sorting.

My version for your whole code:

import datetime as DT

# unsorted dates
raw = ("2010-08-01","2010-03-25","2010-07-01")

datetuples = [tuple(int(numstr) for numstr in dt.split('-'))
              for dt in raw] ## numbers from strings 

transactions = [(DT.date(* dateint),
                 "Some data here") for dateint in datetuples]

transactions.sort()

for info in transactions:
    print info

Answer 3

Well, the reason it's not sorted is that you haven't reassigned the sorted list back to transactions; you want:

transactions = sorted(transactions, key=itemgetter(0))

I should point out that datetime has a strptime function that does what you're doing manually:

transactions = [ ( DT.datetime.strptime( datestring, "%Y-%m-%d" ).date(), "Some data here" ) for datestring in datestrings ]

You might also prefer a collections.OrderedDict, although maybe not...

alt_transactions = collections.OrderedDict( sorted( transactions.items( ) ) )