Itertools are a mighty powerful way to lazily, efficiently work any iterable data structure like lists or strings. For example, imap and ifilter like their plain counter parts map and filter operate on a list, but lazily by returning iterator

Groupby()

Not quite the SQL groupby() you may be used to. Instead groupby accepts an iterable data structure and a mapping function. The goal is to return consecutive items in group. Here’s an example,

from itertools import groupby
writers = [("Tolstoy", "Russia"), ("Dostoyevsky", "Russia"), 
           ("Hemingway", "America"), ("Sartre", "France"), 
           ("Turgenev", "Russia")]

for writer, country in groupby(writers, lambda w: w[1]):
    print(w, c)

returns

>>> ('Russia', <itertools._grouper object at 0x107165e90>)
>>> ('America', <itertools._grouper object at 0x107165ed0>)
>>> ('France', <itertools._grouper object at 0x107165e90>)
>>> ('Russia', <itertools._grouper object at 0x107165ed0>)

The second element of each tuple is an iterable, grouping writers of the same country that appearing one after another. Note “Russia” appears twice, because “Turgenev” while Russian doesn’t appear after or before another Russian writer.

You can store each group in a list for example and use the consecutive grouping for other processing.

This post is a wonderful reference.