#defaultdict

[petaminds]

Advanced python part 3

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…

View On WordPress

[jaakkotulkki]

Python’s collections.defaultdict in Action

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.

l = collections.defaultdict(set) for name, home_city in people: l[name].add(home_city) print(l) # defaultdict(<class 'set'>, {'Mark': {'Chicago'}, 'David': {'St. Louis', 'Miami'}, 'Jane': {'SF', 'Dallas', 'Austin'}}) print(l['chicken']) # set() i = collections.defaultdict(int) for name, city in people: i[name] += 1 print(i) # defaultdict(<class 'int'>, {'Mark': 1, 'David': 2, 'Jane': 3}) print(i['chicken'])

[widu]

Python's defaultdict

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

And you can go crazy with as many nested dicts as you wish!

[makkalot-blog]