Question

I don’t think that I found out a bug but it doesn’t look normal to me.

from itertools import groupby
from operator import itemgetter
c=[((u'http://www.example.com', u'second_value'), u'one'), 
   ((u'http://www.example.com', u'second_value'), u'two'), 
   ((u'http://www.hello.com', u'second_value'), u'one'), 
   ((u'http://www.example.com', u'second_value'), u'three'), 
   ((u'http://www.hello.com', u'second_value'), u'two')]
b= groupby(c, key=itemgetter(0))
for unique_keys, group in b:
    print unique_keys

Yields:

(u'http://www.example.com', u'second_value')
(u'http://www.hello.com', u'second_value')
(u'http://www.example.com', u'second_value')
(u'http://www.hello.com', u'second_value')

Any explanations ? (I was expecting only two different keys). I am using python 2.7.1 if that makes a difference

Answer 1

The iterable needs to already be sorted (on the same key function):

from itertools import groupby
from operator import itemgetter
c=[((u'http://www.example.com', u'second_value'), u'one'), 
   ((u'http://www.example.com', u'second_value'), u'two'), 
   ((u'http://www.hello.com', u'second_value'), u'one'), 
   ((u'http://www.example.com', u'second_value'), u'three'), 
   ((u'http://www.hello.com', u'second_value'), u'two')]
b= groupby(sorted(c,key=itemgetter(0)), key=itemgetter(0))
for unique_keys, group in b:
    print unique_keys

out:

(u'http://www.example.com', u'second_value')
(u'http://www.hello.com', u'second_value')