Related
I am looking to take as input a list and then create another list which contains tuples (or sub-lists) of adjacent elements from the original list, wrapping around for the beginning and ending elements. The input/output would look like this:
l_in = [0, 1, 2, 3]
l_out = [(3, 0, 1), (0, 1, 2), (1, 2, 3), (2, 3, 0)]
My question is closely related to another titled getting successive adjacent elements of a list, but this other question does not take into account wrapping around for the end elements and only handles pairs of elements rather than triplets.
I have a somewhat longer approach to do this involving rotating deques and zipping them together:
from collections import deque
l_in = [0, 1, 2, 3]
deq = deque(l_in)
deq.rotate(1)
deq_prev = deque(deq)
deq.rotate(-2)
deq_next = deque(deq)
deq.rotate(1)
l_out = list(zip(deq_prev, deq, deq_next))
# l_out is [(3, 0, 1), (0, 1, 2), (1, 2, 3), (2, 3, 0)]
However, I feel like there is probably a more elegant (and/or efficient) way to do this using other built-in Python functionality. If, for instance, the rotate() function of deque returned the rotated list instead of modifying it in place, this could be a one- or two-liner (though this approach of zipping together rotated lists is perhaps not the most efficient). How can I accomplish this more elegantly and/or efficiently?
One approach may be to use itertools combined with more_itertools.windowed:
import itertools as it
import more_itertools as mit
l_in = [0, 1, 2, 3]
n = len(l_in)
list(it.islice(mit.windowed(it.cycle(l_in), 3), n-1, 2*n-1))
# [(3, 0, 1), (0, 1, 2), (1, 2, 3), (2, 3, 0)]
Here we generated an infinite cycle of sliding windows and sliced the desired subset.
FWIW, here is an abstraction of the latter code for a general, flexible solution given any iterable input e.g. range(5), "abcde", iter([0, 1, 2, 3]), etc.:
def get_windows(iterable, size=3, offset=-1):
"""Return an iterable of windows including an optional offset."""
it1, it2 = it.tee(iterable)
n = mit.ilen(it1)
return it.islice(mit.windowed(it.cycle(it2), size), n+offset, 2*n+offset)
list(get_windows(l_in))
# [(3, 0, 1), (0, 1, 2), (1, 2, 3), (2, 3, 0)]
list(get_windows("abc", size=2))
# [('c', 'a'), ('a', 'b'), ('b', 'c')]
list(get_windows(range(5), size=2, offset=-2))
# [(3, 4), (4, 0), (0, 1), (1, 2), (2, 3)]
Note: more-itertools is a separate library, easily installed via:
> pip install more_itertools
This can be done with slices:
l_in = [0, 1, 2, 3]
l_in = [l_in[-1]] + l_in + [l_in[0]]
l_out = [l_in[i:i+3] for i in range(len(l_in)-2)]
Well, or such a perversion:
div = len(l_in)
n = 3
l_out = [l_in[i % div: i % div + 3]
if len(l_in[i % div: i % div + 3]) == 3
else l_in[i % div: i % div + 3] + l_in[:3 - len(l_in[i % div: i % div + 3])]
for i in range(3, len(l_in) + 3 * n + 2)]
You can specify the number of iterations.
Well I figured out a better solution as I was writing the question, but I already went through the work of writing it, so here goes. This solution is at least much more concise:
l_out = list(zip(l_in[-1:] + l_in[:-1], l_in, l_in[1:] + l_in[:1]))
See this post for different answers on how to rotate lists in Python.
The one-line solution above should be at least as efficient as the solution in the question (based on my understanding) since the slicing should not be more expensive than the rotating and copying of the deques (see https://wiki.python.org/moin/TimeComplexity).
Other answers with more efficient (or elegant) solutions are still welcome though.
as you found there is a list rotation slicing based idiom lst[i:] + lst[:i]
using it inside a comprehension taking a variable n for the number of adjacent elements wanted is more general [lst[i:] + lst[:i] for i in range(n)]
so everything can be parameterized, the number of adjacent elements n in the cyclic rotation and the 'phase' p, the starting point if not the 'natural' 0 base index, although the default p=-1 is set to -1 to fit the apparant desired output
tst = list(range(4))
def rot(lst, n, p=-1):
return list(zip(*([lst[i+p:] + lst[:i+p] for i in range(n)])))
rot(tst, 3)
Out[2]: [(3, 0, 1), (0, 1, 2), (1, 2, 3), (2, 3, 0)]
showing the shortend code as per the comment
Is there a nice Pythonic way to loop over a list, retuning a pair of elements? The last element should be paired with the first.
So for instance, if I have the list [1, 2, 3], I would like to get the following pairs:
1 - 2
2 - 3
3 - 1
A Pythonic way to access a list pairwise is: zip(L, L[1:]). To connect the last item to the first one:
>>> L = [1, 2, 3]
>>> zip(L, L[1:] + L[:1])
[(1, 2), (2, 3), (3, 1)]
I would use a deque with zip to achieve this.
>>> from collections import deque
>>>
>>> l = [1,2,3]
>>> d = deque(l)
>>> d.rotate(-1)
>>> zip(l, d)
[(1, 2), (2, 3), (3, 1)]
I'd use a slight modification to the pairwise recipe from the itertools documentation:
def pairwise_circle(iterable):
"s -> (s0,s1), (s1,s2), (s2, s3), ... (s<last>,s0)"
a, b = itertools.tee(iterable)
first_value = next(b, None)
return itertools.zip_longest(a, b,fillvalue=first_value)
This will simply keep a reference to the first value and when the second iterator is exhausted, zip_longest will fill the last place with the first value.
(Also note that it works with iterators like generators as well as iterables like lists/tuples.)
Note that #Barry's solution is very similar to this but a bit easier to understand in my opinion and easier to extend beyond one element.
I would pair itertools.cycle with zip:
import itertools
def circular_pairwise(l):
second = itertools.cycle(l)
next(second)
return zip(l, second)
cycle returns an iterable that yields the values of its argument in order, looping from the last value to the first.
We skip the first value, so it starts at position 1 (rather than 0).
Next, we zip it with the original, unmutated list. zip is good, because it stops when any of its argument iterables are exhausted.
Doing it this way avoids the creation of any intermediate lists: cycle holds a reference to the original, but doesn't copy it. zip operates in the same way.
It's important to note that this will break if the input is an iterator, such as a file, (or a map or zip in python-3), as advancing in one place (through next(second)) will automatically advance the iterator in all the others. This is easily solved using itertools.tee, which produces two independently operating iterators over the original iterable:
def circular_pairwise(it):
first, snd = itertools.tee(it)
second = itertools.cycle(snd)
next(second)
return zip(first, second)
tee can use large amounts of additional storage, for example, if one of the returned iterators is used up before the other is touched, but as we only ever have one step difference, the additional storage is minimal.
There are more efficient ways (that don't built temporary lists), but I think this is the most concise:
> l = [1,2,3]
> zip(l, (l+l)[1:])
[(1, 2), (2, 3), (3, 1)]
Pairwise circular Python 'for' loop
If you like the accepted answer,
zip(L, L[1:] + L[:1])
you can go much more memory light with semantically the same code using itertools:
from itertools import islice, chain #, izip as zip # uncomment if Python 2
And this barely materializes anything in memory beyond the original list (assuming the list is relatively large):
zip(l, chain(islice(l, 1, None), islice(l, None, 1)))
To use, just consume (for example, with a list):
>>> list(zip(l, chain(islice(l, 1, None), islice(l, None, 1))))
[(1, 2), (2, 3), (3, 1)]
This can be made extensible to any width:
def cyclical_window(l, width=2):
return zip(*[chain(islice(l, i, None), islice(l, None, i)) for i in range(width)])
and usage:
>>> l = [1, 2, 3, 4, 5]
>>> cyclical_window(l)
<itertools.izip object at 0x112E7D28>
>>> list(cyclical_window(l))
[(1, 2), (2, 3), (3, 4), (4, 5), (5, 1)]
>>> list(cyclical_window(l, 4))
[(1, 2, 3, 4), (2, 3, 4, 5), (3, 4, 5, 1), (4, 5, 1, 2), (5, 1, 2, 3)]
Unlimited generation with itertools.tee with cycle
You can also use tee to avoid making a redundant cycle object:
from itertools import cycle, tee
ic1, ic2 = tee(cycle(l))
next(ic2) # must still queue up the next item
and now:
>>> [(next(ic1), next(ic2)) for _ in range(10)]
[(1, 2), (2, 3), (3, 1), (1, 2), (2, 3), (3, 1), (1, 2), (2, 3), (3, 1), (1, 2)]
This is incredibly efficient, an expected usage of iter with next, and elegant usage of cycle, tee, and zip.
Don't pass cycle directly to list unless you have saved your work and have time for your computer to creep to a halt as you max out its memory - if you're lucky, after a while your OS will kill the process before it crashes your computer.
Pure Python Builtin Functions
Finally, no standard lib imports, but this only works for up to the length of original list (IndexError otherwise.)
>>> [(l[i], l[i - len(l) + 1]) for i in range(len(l))]
[(1, 2), (2, 3), (3, 1)]
You can continue this with modulo:
>>> len_l = len(l)
>>> [(l[i % len_l], l[(i + 1) % len_l]) for i in range(10)]
[(1, 2), (2, 3), (3, 1), (1, 2), (2, 3), (3, 1), (1, 2), (2, 3), (3, 1), (1, 2)]
I would use a list comprehension, and take advantage of the fact that l[-1] is the last element.
>>> l = [1,2,3]
>>> [(l[i-1],l[i]) for i in range(len(l))]
[(3, 1), (1, 2), (2, 3)]
You don't need a temporary list that way.
Amazing how many different ways there are to solve this problem.
Here's one more. You can use the pairwise recipe but instead of zipping with b, chain it with the first element that you already popped off. Don't need to cycle when we just need a single extra value:
from itertools import chain, izip, tee
def pairwise_circle(iterable):
a, b = tee(iterable)
first = next(b, None)
return izip(a, chain(b, (first,)))
I like a solution that does not modify the original list and does not copy the list to temporary storage:
def circular(a_list):
for index in range(len(a_list) - 1):
yield a_list[index], a_list[index + 1]
yield a_list[-1], a_list[0]
for x in circular([1, 2, 3]):
print x
Output:
(1, 2)
(2, 3)
(3, 1)
I can imagine this being used on some very large in-memory data.
This one will work even if the list l has consumed most of the system's memory. (If something guarantees this case to be impossible, then zip as posted by chepner is fine)
l.append( l[0] )
for i in range( len(l)-1):
pair = l[i],l[i+1]
# stuff involving pair
del l[-1]
or more generalizably (works for any offset n i.e. l[ (i+n)%len(l) ] )
for i in range( len(l)):
pair = l[i], l[ (i+1)%len(l) ]
# stuff
provided you are on a system with decently fast modulo division (i.e. not some pea-brained embedded system).
There seems to be a often-held belief that indexing a list with an integer subscript is un-pythonic and best avoided. Why?
This is my solution, and it looks Pythonic enough to me:
l = [1,2,3]
for n,v in enumerate(l):
try:
print(v,l[n+1])
except IndexError:
print(v,l[0])
prints:
1 2
2 3
3 1
The generator function version:
def f(iterable):
for n,v in enumerate(iterable):
try:
yield(v,iterable[n+1])
except IndexError:
yield(v,iterable[0])
>>> list(f([1,2,3]))
[(1, 2), (2, 3), (3, 1)]
How about this?
li = li+[li[0]]
pairwise = [(li[i],li[i+1]) for i in range(len(li)-1)]
from itertools import izip, chain, islice
itr = izip(l, chain(islice(l, 1, None), islice(l, 1)))
(As above with #j-f-sebastian's "zip" answer, but using itertools.)
NB: EDITED given helpful nudge from #200_success. previously was:
itr = izip(l, chain(l[1:], l[:1]))
If you don't want to consume too much memory, you can try my solution:
[(l[i], l[(i+1) % len(l)]) for i, v in enumerate(l)]
It's a little slower, but consume less memory.
Starting in Python 3.10, the new pairwise function provides a way to create sliding pairs of consecutive elements:
from itertools import pairwise
# l = [1, 2, 3]
list(pairwise(l + l[:1]))
# [(1, 2), (2, 3), (3, 1)]
or simply pairwise(l + l[:1]) if you don't need the result as a list.
Note that we pairwise on the list appended with its head (l + l[:1]) so that rolling pairs are circular (i.e. so that we also include the (3, 1) pair):
list(pairwise(l)) # [(1, 2), (2, 3)]
l + l[:1] # [1, 2, 3, 1]
Just another try
>>> L = [1,2,3]
>>> zip(L,L[1:]) + [(L[-1],L[0])]
[(1, 2), (2, 3), (3, 1)]
L = [1, 2, 3]
a = zip(L, L[1:]+L[:1])
for i in a:
b = list(i)
print b
this seems like combinations would do the job.
from itertools import combinations
x=combinations([1,2,3],2)
this would yield a generator. this can then be iterated over as such
for i in x:
print i
the results would look something like
(1, 2)
(1, 3)
(2, 3)
Given a list of items in Python, how can I get all the possible combinations of the items?
There are several similar questions on this site, that suggest using itertools.combinations, but that returns only a subset of what I need:
stuff = [1, 2, 3]
for L in range(0, len(stuff)+1):
for subset in itertools.combinations(stuff, L):
print(subset)
()
(1,)
(2,)
(3,)
(1, 2)
(1, 3)
(2, 3)
(1, 2, 3)
As you see, it returns only items in a strict order, not returning (2, 1), (3, 2), (3, 1), (2, 1, 3), (3, 1, 2), (2, 3, 1), and (3, 2, 1). Is there some workaround for that? I can't seem to come up with anything.
Use itertools.permutations:
>>> import itertools
>>> stuff = [1, 2, 3]
>>> for L in range(0, len(stuff)+1):
for subset in itertools.permutations(stuff, L):
print(subset)
...
()
(1,)
(2,)
(3,)
(1, 2)
(1, 3)
(2, 1)
(2, 3)
(3, 1)
....
Help on itertools.permutations:
permutations(iterable[, r]) --> permutations object
Return successive r-length permutations of elements in the iterable.
permutations(range(3), 2) --> (0,1), (0,2), (1,0), (1,2), (2,0), (2,1)
You can generate all the combinations of a list in python using this simple code
import itertools
a = [1,2,3,4]
for i in xrange(1,len(a)+1):
print list(itertools.combinations(a,i))
Result:
[(1,), (2,), (3,), (4,)]
[(1, 2), (1, 3), (1, 4), (2, 3), (2, 4), (3, 4)]
[(1, 2, 3), (1, 2, 4), (1, 3, 4), (2, 3, 4)]
[(1, 2, 3, 4)]
Are you looking for itertools.permutations instead?
From help(itertools.permutations),
Help on class permutations in module itertools:
class permutations(__builtin__.object)
| permutations(iterable[, r]) --> permutations object
|
| Return successive r-length permutations of elements in the iterable.
|
| permutations(range(3), 2) --> (0,1), (0,2), (1,0), (1,2), (2,0), (2,1)
Sample Code :
>>> from itertools import permutations
>>> stuff = [1, 2, 3]
>>> for i in range(0, len(stuff)+1):
for subset in permutations(stuff, i):
print(subset)
()
(1,)
(2,)
(3,)
(1, 2)
(1, 3)
(2, 1)
(2, 3)
(3, 1)
(3, 2)
(1, 2, 3)
(1, 3, 2)
(2, 1, 3)
(2, 3, 1)
(3, 1, 2)
(3, 2, 1)
From Wikipedia, the difference between permutations and combinations :
Permutation :
Informally, a permutation of a set of objects is an arrangement of those objects into a particular order. For example, there are six permutations of the set {1,2,3}, namely (1,2,3), (1,3,2), (2,1,3), (2,3,1), (3,1,2), and (3,2,1).
Combination :
In mathematics a combination is a way of selecting several things out of a larger group, where (unlike permutations) order does not matter.
itertools.permutations is going to be what you want. By mathematical definition, order does not matter for combinations, meaning (1,2) is considered identical to (2,1). Whereas with permutations, each distinct ordering counts as a unique permutation, so (1,2) and (2,1) are completely different.
Here is a solution without itertools
First lets define a translation between an indicator vector of 0 and 1s and a sub-list (1 if the item is in the sublist)
def indicators2sublist(indicators,arr):
return [item for item,indicator in zip(arr,indicators) if int(indicator)==1]
Next, Well define a mapping from a number between 0 and 2^n-1 to the its binary vector representation (using string's format function) :
def bin(n,sz):
return ('{d:0'+str(sz)+'b}').format(d=n)
All we have left to do, is to iterate all the possible numbers, and call indicators2sublist
def all_sublists(arr):
sz=len(arr)
for n in xrange(0,2**sz):
b=bin(n,sz)
yield indicators2sublist(b,arr)
I assume you want all possible combinations as 'sets' of values. Here is a piece of code that I wrote that might help give you an idea:
def getAllCombinations(object_list):
uniq_objs = set(object_list)
combinations = []
for obj in uniq_objs:
for i in range(0,len(combinations)):
combinations.append(combinations[i].union([obj]))
combinations.append(set([obj]))
return combinations
Here is a sample:
combinations = getAllCombinations([20,10,30])
combinations.sort(key = lambda s: len(s))
print combinations
... [set([10]), set([20]), set([30]), set([10, 20]), set([10, 30]), set([20, 30]), set([10, 20, 30])]
I think this has n! time complexity, so be careful. This works but may not be most efficient
just thought i'd put this out there since i couldn't fine EVERY possible outcome and keeping in mind i only have the rawest most basic of knowledge when it comes to python and there's probably a much more elegant solution...(also excuse the poor variable names
testing = [1, 2, 3]
testing2= [0]
n = -1
def testingSomethingElse(number):
try:
testing2[0:len(testing2)] == testing[0]
n = -1
testing2[number] += 1
except IndexError:
testing2.append(testing[0])
while True:
n += 1
testing2[0] = testing[n]
print(testing2)
if testing2[0] == testing[-1]:
try:
n = -1
testing2[1] += 1
except IndexError:
testing2.append(testing[0])
for i in range(len(testing2)):
if testing2[i] == 4:
testingSomethingElse(i+1)
testing2[i] = testing[0]
i got away with == 4 because i'm working with integers but you may have to modify that accordingly...
I have a list of 46 items. Each has a number associated with it. I want to pair these items up in a set of 23 pairs. I want to evaluate a function over each set. How do I generate such a set?
I can use the combinations function from itertools to produce all the 2-ples but I don't see how to generate all the sets of 23 pairs.
How do I do this or is there sample code I can reference?
>>> L=range(46)
>>> def f(x, y): #for example
... return x * y
...
>>> [f(x, y) for x, y in zip(*[iter(L)] * 2)]
[0, 6, 20, 42, 72, 110, 156, 210, 272, 342, 420, 506, 600, 702, 812, 930, 1056, 1190, 1332, 1482, 1640, 1806, 1980]
Edit:
For the powerset of the pairs, we start by creating the pairs the same way. For Python3 use range in place of xrange
S = zip(*[iter(L)] * 2) # set of 23 pairs
[{j for i, j in enumerate(S) if (1<<i)&k} for k in xrange(1<<len(S))]
This will be quite a big list, you may want to use a generator expression
for item in ({j for i, j in enumerate(S) if (1<<i)&k} for k in xrange(1<<len(S))):
func(item)
First, the natural way to get all the pairs from a list is:
>>> N = 10
>>> input_list = range(N)
>>> [(a,b) for a, b in zip(input_list[::2], input_list[1::2])]
[(0, 1), (2, 3), (4, 5), (6, 7), (8, 9)]
If you want to generate all such pairs, I'd do something like (this is what I call Case 1 below):
>>> set_of_all_pairs = set()
>>> input_list = range(N)
>>> import itertools
>>> for perm in itertools.permutations(input_list):
pairs = tuple([(a,b) for a, b in zip(perm[::2], perm[1::2])])
set_of_all_pairs.add(pairs)
Granted this as is will differentiate order in pair (e.g., (1,4) is different than (4,1)) as well as consider the order of pairs meaningful. So if you sort the pairs and the set of pairs before adding to the set:
>>> set_of_all_pairs = set()
>>> input_list = range(N)
>>> import itertools
>>> for perm in itertools.permutations(input_list):
pairs = sorted([tuple(sorted((a,b))) for a, b in zip(perm[::2], perm[1::2])])
set_of_all_pairs.add(tuple(pairs))
This is not an efficient algorithm (what I call Case 3 below), but for small values of N it will work.
For N=6, using the sorted method.
set([((0, 4), (1, 3), (2, 5)),
((0, 4), (1, 5), (2, 3)),
((0, 1), (2, 3), (4, 5)),
((0, 3), (1, 5), (2, 4)),
((0, 2), (1, 5), (3, 4)),
((0, 4), (1, 2), (3, 5)),
((0, 3), (1, 4), (2, 5)),
((0, 1), (2, 4), (3, 5)),
((0, 5), (1, 4), (2, 3)),
((0, 5), (1, 2), (3, 4)),
((0, 2), (1, 3), (4, 5)),
((0, 3), (1, 2), (4, 5)),
((0, 2), (1, 4), (3, 5)),
((0, 1), (2, 5), (3, 4)),
((0, 5), (1, 3), (2, 4))])
Note the solution space grows exponentially fast; (e.g., for N=6 its 15; N=8 its 105; N=10, its 945, for N=46 will be 25373791335626257947657609375 ~ 2.5 x 1028).
EDIT: People criticized the O(N!), but the desired solution grows as O(N!)
The question asks to break a list of N elements (assuming most general case of all elements being distinct) into a set of (N/2) pairs, and not only do this once, but generate all sets of these pairings. This answer is the only one that does so. Yes, it's exponentially slow, and completely infeasible for N=46. That's why I used N=10.
There are three reasonable interpretations of the problem:
Case 1: Ordering matters both inside a pair in the tuple (e.g., function arguments are not symmetric) and in the order of the pairs in a set of pairs also matters, then we will have N! ways of pairing up the numbers in our answer. Meaning in this case both the pair (0,1) and (1,0) are consider distinct, as well as for the N=4 case we consider the pairings {(0,1), (2,3)} distinct from {(2,3),(0,1)}.
Case 2: Ordering matters in a pair, but order is irrelevant in a set of pairings. This means we consider (0,1) and (1,0) as distinct pairs, but consider (for the N=4 case) that the set {(0,1),(2,3)} is identical to the set {(2,3), (0,1)} and do not need to consider both. In this case we will have N!/(N/2)! pairings, as any given set has (N/2)! different orderings. (I didn't explicitly give this above; but just stop sorting the tuple).
Case 3: Ordering is irrelevant both within a pair and within a set of pairings. This means we consider (0,1) and (1,0) as the same pair (function arguments are symmetric), so we will have N!/( (N/2)! & 2^(N/2) ) sets of pairs (factorial(N)/(factorial(N/2)*2**(N/2))). Each of the (N/2) pairs in each combination will have two internal orderings that contribute.
So depending on how the problem is phrased we should have:
Case 1 | Case 2 | Case 3
----------------------------------------------
N N! | N!/(N/2)! | N!/((N/2)! 2^(N/2))
6 720 | 120 | 15
8 40320 | 1680 | 105
10 3628800 | 30240 | 945
46 5.5x10^57 | 2.1x10^35 | 2x10^28
Note, my algorithm will go through all permutations, and hence will actually run slower for Case 3 (due to sorting) than Case 1, even though a better algorithm for Case 3 could be much faster. However, my answer is still optimal in asymptotic notation as even case 3 is factorial in its asymptotic running time, and completely infeasible to solve for N~46. Granted if you had to do a problem-size at the limit of feasibility (N~16) for Case 3 (e.g., need to generate 518918400.0 pairings), this solution of iterating through all N! permutations, sorting, and throwing out duplicates is sub-optimal.
Hi:
I'm trying to sort a list of tuples in a custom way:
For example:
lt = [(2,4), (4,5), (5,2)]
must be sorted:
lt = [(5,2), (2,4), (4,5)]
Rules:
* b tuple is greater than a tuple if a[1] == b[0]
* a tuple is greater than b tuple if a[0] == b[1]
I've implemented a cmp function like this:
def tcmp(a, b):
if a[1] == b[0]:
return -1
elif a[0] == b[1]:
return 1
else:
return 0
but sorting the list:
lt.sort(tcmp)
lt show me:
lt = [(2, 4), (4, 5), (5, 2)]
What am I doing wrong?
Sounds a lot to me you are trying to solve one of the Google's Python class problems, which is to sort a list of tuples in increasing order based on their last element.
This how I did it:
def sort_last(tuples):
def last_value_tuple(t):
return t[-1]
return sorted(tuples, key=last_value_tuple)
EDIT: I didn't read the whole thing, and I assumed it was based on the last element of the tuple. Well, still I'm going to leave it here because it can be useful to anyone.
You could also write your code using lambda
def sort(tuples):
return sorted (tuples,key=lambda last : last[-1])
so sort([(1, 3), (3, 2), (2, 1)]) would yield [(2, 1), (3, 2), (1, 3)]
You can write your own custom key function to specify the key value for sorting.
Ex.
def sort_last(tuples):
return sorted(tuples, key=last)
def last(a):
return a[-1]
tuples => sorted tuple by last element
[(1, 3), (3, 2), (2, 1)] => [(2, 1), (3, 2), (1, 3)]
[(1, 7), (1, 3), (3, 4, 5), (2, 2)] => [(2, 2), (1, 3), (3, 4, 5), (1, 7)]
I'm not sure your comparison function is a valid one in a mathematical sense, i.e. transitive. Given a, b, c a comparison function saying that a > b and b > c implies that a > c. Sorting procedures rely on this property.
Not to mention that by your rules, for a = [1, 2] and b = [2, 1] you have both a[1] == b[0] and a[0] == b[1] which means that a is both greater and smaller than b.
Your ordering specification is wrong because it is not transitive.
Transitivity means that if a < b and b < c, then a < c. However, in your case:
(1,2) < (2,3)
(2,3) < (3,1)
(3,1) < (1,2)
Try lt.sort(tcmp, reverse=True).
(While this may produce the right answer, there may be other problems with your comparison method)