Dictionary error in Python 2.7 - python-2.7

Question

I have a file in the format:
0000 | a1_1,a3_2 | b2_1, b3_2
0001 | a1_3 | b4_1
and I'm trying to create a dictionary which has
{ 'a1' : set(['b2', 'b3', 'b4']), 'a3': set(['b2', 'b3']) }
and this is how my code looks like:
def get_ids(row, col):
ids = set()
x = row.strip().split('|')
for a in x[col].split(','):
ids.add(a.split('_')[0])
return ids
def add_to_dictionary(funky_dictionary,key, values):
if key in funky_dictionary:
funky_dictionary[key].update(values)
else:
funky_dictionary[key] = values
def get_dict(input_file):
funky_dictionary = {}
with open(input_file,'r') as ip:
for row in ip:
a_ids = get_ids(row,1)
b_ids = get_ids(row,2)
for key in a_ids:
add_to_dictionary(funky_dictionary,key,b_ids)
return funky_dictionary
So my problem is this when I lookup values for certain key in the dictionary, it returns me with way more values than expected. E.g.
For the above example the expected value of a3 would be set(['b2', ' b3'])
However with the code, I'm getting set(['b2', ' b3', 'b4'])
I cant figure out whats wrong with the code. Any help?

Answer 1

The issue you have is that many of your dictionary's values are in fact references to the same set instances. In your example data, when the first line is processed, 'a1' and 'a3' both get mapped to the same set object (containing 'b2' and 'b3'). When you process the second line and call update on that set via the key 'a1', you'll see the added value through 'a3' too, since both values are references to the same set.
You need to change the code so that each value is a separate set object. I'd suggest getting rid of add_to_dictionary and just using the dictionary's own setdefault method, like this:
for key in a_ids:
funky_dictionary.setdefault(key, set()).update(b_ids)
This code always starts with a new empty set for a new key, and always updates it with new values (rather than adding a reference to the b_ids set to the dictionary directly).