def test_return_dataset_of_size_n(self):
subset = RandomPicker().get_random_subset(
Dataset(Apk('apk1'), Apk('apk2'), Apk('apk3')), 2)
self.assertEqual(len(subset), 2)
subset = RandomPicker().get_random_subset(
Dataset(Apk('apk1'), Apk('apk2'), Apk('apk3'), Apk('apk4')), 3)
self.assertEqual(len(subset), 3)
def test_return_subset_of_dataset(self):
initial_dataset = Dataset(Apk('apk1'), Apk('apk2'), Apk('apk3'),
Apk('apk4'), Apk('apk5'), Apk('apk6'))
subset = RandomPicker().get_random_subset(initial_dataset, 4)
self.assertTrue(initial_dataset.contains(subset))
subset = RandomPicker().get_random_subset(initial_dataset, 2)
self.assertTrue(initial_dataset.contains(subset))
self.assertEqual(len(subset), 2)
def test_return_random_subset(self):
initial_dataset = Dataset(Apk('apk1'), Apk('apk2'), Apk('apk3'),
Apk('apk4'), Apk('apk5'), Apk('apk6'))
picker = RandomPicker(seed=12)
random_subset1 = picker.get_random_subset(initial_dataset, 3)
expected_random_subset1 = Dataset(Apk('apk3'), Apk('apk4'),
Apk('apk1'))
self.assertEqual(random_subset1, expected_random_subset1)
random_subset2 = picker.get_random_subset(initial_dataset, 3)
expected_random_subset2 = Dataset(Apk('apk2'), Apk('apk5'),
Apk('apk6'))
self.assertEqual(random_subset2, expected_random_subset2)
random_subset3 = picker.get_random_subset(initial_dataset, 3)
expected_random_subset3 = Dataset(Apk('apk1'), Apk('apk4'),
Apk('apk3'))
self.assertEqual(random_subset3, expected_random_subset3)
def setUp(self):
self.apk = Apk(sha256='1234', pkg_name='apk1')
self.apk2 = Apk(sha256='5678', pkg_name='apk2')
self.dataset = Dataset(self.apk, self.apk2)
self.constructor_mock = mock.create_autospec(UrlConstructor)
self.constructor_mock.construct.side_effect = lambda apk: ' https://' + apk.sha256 # use sha256 of apk as download url, just for a test
self.dataset_downloader = DatasetDownloader(
self.dataset, url_constructor=self.constructor_mock, out_dir='out')
def test_saves_metadata(self):
dataset = Dataset(Apk(pkg_name='apk1', apk_size=8, dex_date='01-01-2001', markets='play'), Apk(pkg_name='apk2', apk_size=13, dex_date='01-03-2001', markets='play|china'))
MetadataSaver(dataset=dataset, out_dir='out').save(['pkg_name', 'apk_size', 'dex_date', 'markets'])
expected_out = 'pkg_name,apk_size,dex_date,markets\n\
apk1,8,01-01-2001,play\n\
apk2,13,01-03-2001,china|play\n\
'
self.assertTrue(os.path.exists(r'out/metadata.csv'))
with open('out/metadata.csv') as f:
self.assertEqual(expected_out, f.read())
def get_random_subset(self, input_dataset, requested_size):
result = []
for num, apk in enumerate(input_dataset, 1):
if requested_size is DownloadType.ALL or len(
result) < requested_size:
result.append(apk)
else:
s = int(random.random() * num)
if s < requested_size:
result[s] = apk
return Dataset(*result)
def test_saves_metadata(self):
dataset = Dataset(
Apk(pkg_name='apk1',
apk_size=8,
dex_date='01-01-2001',
markets='play'),
Apk(pkg_name='apk2',
apk_size=13,
dex_date='01-03-2001',
markets='play|china'))
MetadataSaver(dataset=dataset,
out_dir='out').save(['pkg_name', 'apk_size', 'dex_date'])
self.assertTrue(os.path.exists(r'out\metadata.csv'))
def test_equals(self):
self.assertEqual(Dataset(), Dataset())
dataset2 = Dataset(Apk('apk1'), Apk('apk2'), Apk('apk3'), Apk('apk4'),
Apk('apk5'))
self.assertEqual(self.dataset, dataset2)
self.assertEqual(hash(self.dataset), hash(dataset2))
dataset2.add(Apk('apk6'))
self.assertNotEqual(hash(self.dataset), hash(dataset2))
def test_returns_whole_dataset_if_dataset_size_less_then_requested_size(
self):
input_dataset = Dataset(Apk('apk1'), Apk('apk2'))
subset = RandomPicker().get_random_subset(input_dataset, 4)
self.assertEqual(subset, input_dataset)
def setUp(self):
self.dataset = Dataset(Apk('apk1'), Apk('apk2'), Apk('apk3'),
Apk('apk4'), Apk('apk5'))