Python Iterate Dictionary by Index

You can iterate over keys and get values by keys:

for key in dict.iterkeys():
    print key, dict[key]

You can iterate over keys and corresponding values:

for key, value in dict.iteritems():
    print key, value

You can use enumerate if you want indexes (remember that dictionaries don't have an order):

>>> for index, key in enumerate(dict):
...     print index, key
... 
0 orange
1 mango
2 apple
>>> 

There are some very good answers here. I'd like to add the following here as well:

some_dict = {
    "foo": "bar",
    "lorem": "ipsum"
}

for index, (key, value) in enumerate(some_dict.items()):
    print(index, key, value)

results in

0 foo bar
1 lorem ipsum

Appears to work with Python 2.7 and 3.5


I wanted to know (idx, key, value) for a python OrderedDict today (mapping of SKUs to quantities in order of the way they should appear on a receipt). The answers here were all bummers.

In python 3, at least, this way works and and makes sense.

In [1]: from collections import OrderedDict
   ...: od = OrderedDict()
   ...: od['a']='spam'
   ...: od['b']='ham'
   ...: od['c']='eggs'
   ...: 
   ...: for i,(k,v) in enumerate(od.items()):
   ...:    print('%d,%s,%s'%(i,k,v))
   ...: 
0,a,spam
1,b,ham
2,c,eggs

Some of the comments are right in saying that these answers do not correspond to the question.

One reason one might want to loop through a dictionary using "indexes" is for example to compute a distance matrix for a set of objects in a dictionary. To put it as an example (going a bit to the basics on the bullet below):

  • Assuming one have 1000 objects on a dictionary, the distance square matrix consider all combinations from one object to any other and so it would have dimensions of 1000x1000 elements. But if the distance from object 1 to object 2 is the same as from object 2 to object 1, one need to compute the distance only to less than half of the square matrix, since the diagonal will have distance 0 and the values are mirrored above and below the diagonal.

This is why most packages use a condensed distance matrix ( How does condensed distance matrix work? (pdist) )

But consider the case one is implementing the computation of a distance matrix, or any kind of permutation of the sort. In such case you need to skip the results from more than half of the cases. This means that a FOR loop that runs through all the dictionary is just hitting an IF and jumping to the next iteration without performing really any job most of the time. For large datasets this additional "IFs" and loops add up to a relevant amount on the processing time and could be avoided if, at each loop, one starts one "index" further on the dictionary.

Going than to the question, my conclusion right now is that the answer is NO. One has no way to directly access the dictionary values by any index except the key or an iterator.

I understand that most of the answers up to now applies different approaches to perform this task but really don't allow any index manipulation, that would be useful in a case such as exemplified.

The only alternative I see is to use a list or other variable as a sequential index to the dictionary. Here than goes an implementation to exemplify such case:

#!/usr/bin/python3

dishes = {'spam': 4.25, 'eggs': 1.50, 'sausage': 1.75, 'bacon': 2.00}
print("Dictionary: {}\n".format(dishes))

key_list = list(dishes.keys())
number_of_items = len(key_list)

condensed_matrix = [0]*int(round(((number_of_items**2)-number_of_items)/2,0))
c_m_index = 0

for first_index in range(0,number_of_items):
    for second_index in range(first_index+1,number_of_items):
        condensed_matrix[c_m_index] = dishes[key_list[first_index]] - dishes[key_list[second_index]]
        print("{}. {}-{} = {}".format(c_m_index,key_list[first_index],key_list[second_index],condensed_matrix[c_m_index]))
        c_m_index+=1

The output is:

Dictionary: {'spam': 4.25, 'eggs': 1.5, 'sausage': 1.75, 'bacon': 2.0}

0. spam-eggs = 2.75
1. spam-sausage = 2.5
2. spam-bacon = 2.25
3. eggs-sausage = -0.25
4. eggs-bacon = -0.5
5. sausage-bacon = -0.25

Its also worth mentioning that are packages such as intertools that allows one to perform similar tasks in a shorter format.