Sort a multidimensional list by a variable number of keys

Question

I've read this post and is hasn't ended up working for me.

Edit: the functionality I'm describing is just like the sorting function in Excel... if that makes it any clearer

Here's my situation, I have a tab-delimited text document. There are about 125,000 lines and 6 columns per line (columns are separated by a tab character). I've split the document into a two-dimension list.

I am trying to write a generic function to sort two-dimensional lists. Basically I would like to have a function where I can pass the big list, and the key of one or more columns I would like to sort the big list by. Obviously, I would like the first key passed to be the primary sorting point, then the second key, etc.

Still confuzzled?

Here's an example of what I would like to do.

Joel    18  Orange  1
Anna    17  Blue    2
Ryan    18  Green   3
Luke    16  Blue    1
Katy    13  Pink    5
Tyler   22  Blue    6
Bob     22  Blue    10
Garrett 24  Red 7
Ryan    18  Green   8
Leland  18  Yellow  9

Say I passed this list to my magical function, like so:

sortByColumn(bigList, 0)

Anna    17  Blue    2
Bob     22  Blue    10
Garrett 24  Red 7
Joel    18  Orange  1
Katy    13  Pink    5
Leland  18  Yellow  9
Luke    16  Blue    1
Ryan    18  Green   3
Ryan    18  Green   8
Tyler   22  Blue    6

and...

sortByColumn(bigList, 2, 3)

Luke    16  Blue    1
Anna    17  Blue    2
Tyler   22  Blue    6
Bob     22  Blue    10
Ryan    18  Green   3
Ryan    18  Green   8
Joel    18  Orange  1
Katy    13  Pink    5
Garrett 24  Red 7
Leland  18  Yellow  9

Any clues?

Answer 1

import operator:
def sortByColumn(bigList, *args)
    bigList.sort(key=operator.itemgetter(*args)) # sorts the list in place

Answer 2

This will sort by columns 2 and 3:

a.sort(key=operator.itemgetter(2,3))

Answer 3

The key idea here (pun intended) is to use a key function that returns a tuple. Below, the key function is lambda x: (x[idx] for idx in args) x is set to equal an element of aList -- that is, a row of data. It returns a tuple of values, not just one value. The sort() method sorts according to the first element of the list, then breaks ties with the second, and so on. See http://wiki.python.org/moin/HowTo/Sorting#Sortingbykeys

#!/usr/bin/env python
import csv
def sortByColumn(aList,*args):
    aList.sort(key=lambda x: (x[idx] for idx in args))
    return aList

filename='file.txt'
def convert_ints(astr):
    try:
        return int(astr)
    except ValueError:
        return astr    
biglist=[[convert_ints(elt) for elt in line]
         for line in csv.reader(open(filename,'r'),delimiter='\t')]

for row in sortByColumn(biglist,0):
    print row

for row in sortByColumn(biglist,2,3):
    print row

Answer 4

Make sure you have converted the numbers to ints, otherwise they will sort alphabetically rather than numerically

# Sort the list in place
def sortByColumn(A,*args):
    import operator
    A.sort(key=operator.itemgetter(*args))
    return A

or

# Leave the original list alone and return a new sorted one
def sortByColumn(A,*args):
    import opertator
    return sorted(A,key=operator.itemgetter(*args))