def _get_similar_files(self, fo_list, exclusive_files):
similars = list()
similarity = dict()
for index, _ in enumerate(fo_list):
tmp_list = deepcopy(fo_list)
parent_one = tmp_list.pop(index)
for parent_two in tmp_list:
for file_one in exclusive_files[parent_one.uid]:
for item, value in self._find_similar_file_for(file=file_one, parent_id=parent_one.uid, potential_matches=parent_two):
similars.append(item)
similarity[convert_uid_list_to_compare_id(item)] = value
similarity_sets = self.produce_similarity_sets(remove_duplicates_from_list_of_lists(similars))
remove_subsets_from_list_of_sets(similarity_sets)
return remove_duplicates_from_list_of_lists(list_of_sets_to_list_of_lists(similarity_sets)), similarity
def test_remove_duplicates_from_list_of_lists(self):
l1 = [[2, 4], [5, 9], [2, 5], [6, 7], [1, 3, 12], [4, 2], [1, 12, 3], [2, 4], [2, 4], [2, 4]]
l2 = [[2, 4], [5, 9], [2, 5], [6, 7], [1, 3, 12]]
self.assertEqual(remove_duplicates_from_list_of_lists(l1), l2, 'result not correct')