itertools.groupby
can do what you want:
import itertools
import operator
L = [('grape', 100), ('grape', 3), ('apple', 15), ('apple', 10),
('apple', 4), ('banana', 3)]
def accumulate(l):
it = itertools.groupby(l, operator.itemgetter(0))
for key, subiter in it:
yield key, sum(item[1] for item in subiter)
print(list(accumulate(L)))
# [('grape', 103), ('apple', 29), ('banana', 3)]