def test_unsupported_version(self):
ts = msprime.simulate(10)
with pytest.raises(ValueError):
tskit.dump_legacy(ts, self.temp_file, version=4)
# Cannot read current files.
ts.dump(self.temp_file)
# Catch Exception here because h5py throws different exceptions on py2 and py3
with pytest.raises(Exception):
tskit.load_legacy(self.temp_file)
def verify_malformed_json_v2(self, ts, group_name, attr, bad_json):
tskit.dump_legacy(ts, self.temp_file, 2)
# Write some bad JSON to the provenance string.
root = h5py.File(self.temp_file, "r+")
group = root[group_name]
group.attrs[attr] = bad_json
root.close()
tsp = tskit.load_legacy(self.temp_file)
self.verify_tree_sequences_equal(ts, tsp)
def test_v2_non_binary_records(self):
demographic_events = [
msprime.SimpleBottleneck(time=0.01, population=0, proportion=1)
]
ts = msprime.simulate(sample_size=10,
demographic_events=demographic_events,
random_seed=1)
with pytest.raises(ValueError):
tskit.dump_legacy(ts, self.temp_file, 2)
def verify_round_trip(self, ts, version):
tskit.dump_legacy(ts, self.temp_file, version=version)
tsp = tskit.load_legacy(self.temp_file)
simplify = version < 10
self.verify_tree_sequences_equal(ts, tsp, simplify=simplify)
tsp.dump(self.temp_file)
tsp = tskit.load(self.temp_file)
self.verify_tree_sequences_equal(ts, tsp, simplify=simplify)
for provenance in tsp.provenances():
tskit.validate_provenance(json.loads(provenance.record))
def test_no_h5py(self):
ts = msprime.simulate(10)
path = os.path.join(test_data_dir, "hdf5-formats", "msprime-0.3.0_v2.0.hdf5")
msg = (
"Legacy formats require h5py. Install via `pip install h5py` or"
" `conda install h5py`"
)
with mock.patch.dict(sys.modules, {"h5py": None}):
with pytest.raises(ImportError, match=msg):
tskit.load_legacy(path)
with pytest.raises(ImportError, match=msg):
tskit.dump_legacy(ts, path)
def test_duplicate_positions_error(self):
ts = msprime.simulate(10, mutation_rate=10)
for version in [2, 3]:
tskit.dump_legacy(ts, self.legacy_file_name, version=version)
root = h5py.File(self.legacy_file_name, "r+")
root['mutations/position'][:] = 0
root.close()
with mock.patch("sys.exit", side_effect=TestException) as mocked_exit:
with self.assertRaises(TestException):
capture_output(
cli.tskit_main,
["upgrade", self.legacy_file_name, self.current_file_name])
self.assertEqual(mocked_exit.call_count, 1)
def test_duplicate_mutation_positions_single_value(self):
ts = multi_locus_with_mutation_example()
for version in [2, 3]:
tskit.dump_legacy(ts, self.temp_file, version=version)
root = h5py.File(self.temp_file, "r+")
root["mutations/position"][:] = 0
root.close()
with pytest.raises(tskit.DuplicatePositionsError):
tskit.load_legacy(self.temp_file)
tsp = tskit.load_legacy(self.temp_file, remove_duplicate_positions=True)
assert tsp.num_sites == 1
sites = list(tsp.sites())
assert sites[0].position == 0
def test_no_h5py(self):
ts = msprime.simulate(10)
msg = ("Legacy formats require h5py. Install via `pip install h5py` or"
" `conda install h5py`")
with h5py.File(self.temp_file, "w") as root:
root["x"] = np.zeros(10)
with mock.patch.dict(sys.modules, {"h5py": None}):
with pytest.raises(ImportError, match=msg):
tskit.load(self.temp_file)
with pytest.raises(ImportError, match=msg):
tskit.load_legacy(self.temp_file)
with pytest.raises(ImportError, match=msg):
tskit.dump_legacy(ts, self.temp_file)
def test_duplicate_positions(self):
ts = msprime.simulate(10, mutation_rate=10)
for version in [2, 3]:
tskit.dump_legacy(ts, self.legacy_file_name, version=version)
root = h5py.File(self.legacy_file_name, "r+")
root['mutations/position'][:] = 0
root.close()
stdout, stderr = capture_output(
cli.tskit_main,
["upgrade", "-d", self.legacy_file_name, self.current_file_name])
self.assertEqual(stdout, "")
tsp = tskit.load(self.current_file_name)
self.assertEqual(tsp.sample_size, ts.sample_size)
self.assertEqual(tsp.num_sites, 1)
def test_conversion(self):
ts1 = msprime.simulate(10)
for version in [2, 3]:
tskit.dump_legacy(ts1, self.legacy_file_name, version=version)
stdout, stderr = capture_output(
cli.tskit_main, [
"upgrade", self.legacy_file_name, self.current_file_name])
ts2 = tskit.load(self.current_file_name)
self.assertEqual(stdout, "")
self.assertEqual(stderr, "")
# Quick checks to ensure we have the right tree sequence.
# More thorough checks are done elsewhere.
self.assertEqual(ts1.get_sample_size(), ts2.get_sample_size())
self.assertEqual(ts1.num_edges, ts2.num_edges)
self.assertEqual(ts1.get_num_trees(), ts2.get_num_trees())
def test_duplicate_mutation_positions(self):
ts = multi_locus_with_mutation_example()
for version in [2, 3]:
tskit.dump_legacy(ts, self.temp_file, version=version)
root = h5py.File(self.temp_file, "r+")
position = np.array(root["mutations/position"])
position[0] = position[1]
root["mutations/position"][:] = position
root.close()
with pytest.raises(tskit.DuplicatePositionsError):
tskit.load_legacy(self.temp_file)
tsp = tskit.load_legacy(self.temp_file, remove_duplicate_positions=True)
assert tsp.num_sites == position.shape[0] - 1
position_after = list(s.position for s in tsp.sites())
assert list(position[1:]) == position_after
def test_general_mutation_example(self):
ts = general_mutation_example()
for version in [2, 3]:
with pytest.raises(ValueError):
tskit.dump_legacy(ts, self.temp_file, version)
self.verify_round_trip(ts, 10)