Start with an iterable which needs to be grouped
lst = [("a", 5, 6), ("b", 2, 4), ("a", 2, 5), ("c", 2, 6)]
Generate the grouped generator, grouping by the second element in each tuple:
def testGroupBy(lst):
groups = itertools.groupby(lst, key=lambda x: x[1])
for key, group in groups:
print(key, list(group))
testGroupBy(lst)
# 5 [('a', 5, 6)]
# 2 [('b', 2, 4), ('a', 2, 5), ('c', 2, 6)]
Only groups of consecutive elements are grouped. You may need to sort by the same key before calling groupby For E.g, (Last element is changed)
lst = [("a", 5, 6), ("b", 2, 4), ("a", 2, 5), ("c", 5, 6)]
testGroupBy(lst)
# 5 [('a', 5, 6)]
# 2 [('b', 2, 4), ('a', 2, 5)]
# 5 [('c', 5, 6)]
The group returned by groupby is an iterator that will be invalid before next iteration. E.g the following will not work if you want the groups to be sorted by key. Group 5 is empty below because when group 2 is fetched it invalidates 5
lst = [("a", 5, 6), ("b", 2, 4), ("a", 2, 5), ("c", 2, 6)]
groups = itertools.groupby(lst, key=lambda x: x[1])
for key, group in sorted(groups):
print(key, list(group))
# 2 [('c', 2, 6)]
# 5 []
To correctly do sorting, create a list from the iterator before sorting
groups = itertools.groupby(lst, key=lambda x: x[1])
for key, group in sorted((key, list(group)) for key, group in groups):
print(key, list(group))
# 2 [('b', 2, 4), ('a', 2, 5), ('c', 2, 6)]
# 5 [('a', 5, 6)]