#defaultdict

Advanced python part 3

In part 2, we introduced advanced python knowledge with built-in functions and other useful tools for sequence iteration, data transformation. Advanced python part 3 will continue with collections to manipulate our data.

Named tuple

Suppose we want to define a data structure to represent a geometric point on a typical x and y-axis. We could easily do this by defining a regular tuple with…

Let's have a look on Pythons collections module that is in Python's standard library. According to the docs, collections module "implements specialized container datatypes providing alternatives to Python’s general purpose built-in containers".

One of the most useful bits of collecions is the defaultdict which is a very useful dict subclass that makes your life a lot easier. Say you are going through a list of people and their hometowns, and you are interested in finding out what kind of home towns each name has (very interesting!).

Using a normal dictionary you might do something along the following lines:

people = [ ('Jane', "Austin",), ('Jane', "SF",), ('Jane', 'Dallas',), ('Mark', "Chicago",), ('David', "St. Louis",), ('David', "Miami") ] normal_dict = {} for name, home_city in people: if name not in normal_dict: normal_dict[name] = [home_city] else: normal_dict[name].append(home_city) print(normal_dict) # {'David': ['St. Louis', 'Miami'], 'Mark': ['Chicago'], 'Jane': ['Austin', 'SF', 'Dallas']} assert('chicken' not in normal_dict)

With defaultdict from collections module you could do it the following way:

import collections l = collections.defaultdict(list) for name, home_city in people: l[name].append(home_city) print(l) # defaultdict(<class 'list'>, {'Jane': ['Austin', 'SF', 'Dallas'], 'David': ['St. Louis', 'Miami'], 'Mark': ['Chicago']}) print(l['chicken']) # []

If you used a normal dictionary and tried appending to a key that does not exist in the dictionary, Python would raise a KeyError, but when defaultdict does not find the key it creates the key automatically. You can also use defaultdict with sets and ints.

The collection lib has many cool classes, like defaultdict. Basically it is a dict that you can initialize with some value, so you don’t need to check if the key exists before each assignment, such as the following annoying computation of histogram: d={}

if item not in d:    

  d[item]=0 d[item]+=1

can be easily replaced with: from collections import defaultdict d=defaultdict(int) d[item]+=1

The cool thing is that you can give arbitrary “default factories”, that are simply a function without parameters. So let’s say you want to have a dict, with each key holding a list of three elements, all initialized to zero: d[k]=[0,0,0] d=defaultdict(lambda: [0,0,0]) d["hello"][1]+=1

Or you can even have a nested defaultdict, if you want double indirection: d = defaultdict(lambda:defaultdict(int)) d["hello"]["world"]+=1