def test_estimate_size_with_sampling_same_size(self):
num_files = 2 * FileBasedSource.MIN_NUMBER_OF_FILES_TO_STAT
pattern, _ = write_pattern([10] * num_files)
# Each line will be of length 6 since write_pattern() uses
# ('line' + line number + '\n') as data.
self.assertEqual(6 * 10 * num_files,
FileBasedSource(pattern).estimate_size())
def test_string_or_value_provider_only(self):
str_file_pattern = tempfile.NamedTemporaryFile(delete=False).name
self.assertEqual(str_file_pattern,
FileBasedSource(str_file_pattern)._pattern.value)
static_vp_file_pattern = StaticValueProvider(value_type=str,
value=str_file_pattern)
self.assertEqual(static_vp_file_pattern,
FileBasedSource(static_vp_file_pattern)._pattern)
runtime_vp_file_pattern = RuntimeValueProvider(
option_name='arg', value_type=str, default_value=str_file_pattern)
self.assertEqual(runtime_vp_file_pattern,
FileBasedSource(runtime_vp_file_pattern)._pattern)
invalid_file_pattern = 123
with self.assertRaises(TypeError):
FileBasedSource(invalid_file_pattern)
def test_estimate_size_with_sampling_different_sizes(self):
num_files = 2 * FileBasedSource.MIN_NUMBER_OF_FILES_TO_STAT
# Each line will be of length 8 since write_pattern() uses
# ('line' + line number + '\n') as data.
base_size = 500
variance = 5
sizes = []
for _ in xrange(num_files):
sizes.append(
int(random.uniform(base_size - variance,
base_size + variance)))
pattern, _ = write_pattern(sizes)
tolerance = 0.05
self.assertAlmostEqual(base_size * 8 * num_files,
FileBasedSource(pattern).estimate_size(),
delta=base_size * 8 * num_files * tolerance)