I have been using defaultdict quite a lot as an substitute to dict for a while, and Counter as well. They both come from python’s standard module collections, which was first introduced in 2.4. Since I am very satisfied with the two, I thought to myself why not check out the rest of the container types in the module. So here I compile this post on python’s collections module, the “High-performance container datatypes” as claimed.

This post will briefly introduce what are the container datatypes in collections and maybe explain why there are of high performance.

As of 2.7, there are now five useful things in collections: namedtuple(), deque, Counter, OrderedDict, defaultdict.

• Counter
  In a nutshell, it helps you do counting; value and its counting are in the form of key value pair.
  Note that Counter is a subclass of dict, so that should clarify a bit on how countings are represented.

  notes:

  • There are three ways to initialize a Counter object: Counter(), Counter({"k1":3, "k2":4}), Counter(k1=3, k2=4).
  • Counter has a dictionary-like interface except that it return zero for a non-existing key instead of a key error.

  • elements() - return an iterator repeating each key as many times as its count.
    In comparison, keys() return a list with all keys in it.

    In [17]: from collections import Counter
      
    In [18]: c = Counter(ad=1, adas=3, adqw=23)
        
    In [19]: c.elements()
    Out[19]: <itertools.chain at 0x10f97b8d0>
        
    In [20]: for i in c.elements():
        ...:     print i
        ...:
    adas
    adas
    adas
    ad
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
    adqw
        
    In [22]: c.keys()
    Out[22]: ['adas', 'ad', 'adqw']
        
    In [23]:
    

  • Since Counter is a subclass of dict, it jas access to keys() to get a list with all keys in itself.
  • most_common([n]) - return n most common keys and their counts in the form of tuples in the counter.
  • subtract([iterable-or-mapping]) - it subtracts counts passed in from current counter. Counts in the result can be zero or negative.
    In comparison, update([iterable-or-mapping]) adds the couts passed in on top of existin counts. Both of them does not replace, but subtracting or adding.
  • We can also do mathmatical operations on counter object: +, -, &, |.

• (TODO: 10000tb) more notes on rest of the container types within collections.

Reference(s):

1. https://docs.python.org/2/library/collections.html