def test_generate_mixes_using_all_examples_plus_random_examples(self):
sourceid_to_exids = ([('source1', i)
for i in range(100)] + [('source2', i)
for i in range(100)])
exid_to_mixids = create_dataset_lib.generate_mixes(
val='unused', num_mixes=200, sourceid_to_exids=sourceid_to_exids)
self.assertEqual(set(range(100)), exid_to_mixids.keys())
def test_generate_unique_mixes(self):
sourceid_to_exids = [('source1', 'a'), ('source1', 'b'),
('source2', 'c'), ('source2', 'd')]
exid_to_mixids = create_dataset_lib.generate_mixes(
val='unused', num_mixes=100, sourceid_to_exids=sourceid_to_exids)
mix_ids = set(itertools.chain(*exid_to_mixids.values()))
# Requested 100, but there are only 4 unique mixes, so that's how many
# we should end up with.
self.assertEqual(4, len(mix_ids))
def test_generate_num_mixes(self):
sourceid_to_exids = [('source1', 'a'), ('source1', 'b'), ('source1', 'c'),
('source2', 'd'), ('source2', 'e'), ('source2', 'f')]
exid_to_mixids = create_dataset_lib.generate_mixes(
val='unused', num_mixes=4, sourceid_to_exids=sourceid_to_exids)
mix_ids = set(itertools.chain(*list(exid_to_mixids.values())))
# Ensure we get the number of mixes we requested even when more unique mixes
# would be possible.
self.assertEqual(4, len(mix_ids))
def test_unique_mixes_duplicate_sources(self):
sourceid_to_exids = [('source1', 'a'), ('source1', 'b'), ('source1', 'c'),
('source2', 'a'), ('source2', 'b'), ('source2', 'c'),
('source3', 'a'), ('source3', 'b'), ('source3', 'c')]
exid_to_mixids = create_dataset_lib.generate_mixes(
val='unused', num_mixes=100, sourceid_to_exids=sourceid_to_exids)
mix_ids = set(itertools.chain(*list(exid_to_mixids.values())))
# There are only 3 unique ids, but we're request mixes of 3 items, so only
# 1 unique mix is possible.
self.assertEqual(1, len(mix_ids))
