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 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_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_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_v2_no_samples(self):
ts = multi_locus_with_mutation_example()
tskit.dump_legacy(ts, self.temp_file, version=2)
root = h5py.File(self.temp_file, "r+")
del root["samples"]
root.close()
tsp = tskit.load_legacy(self.temp_file)
self.verify_tree_sequences_equal(ts, tsp)
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 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 run_upgrade(args):
try:
tree_sequence = tskit.load_legacy(args.source, args.remove_duplicate_positions)
except tskit.DuplicatePositionsError:
exit(
"Error: Duplicate mutation positions in the source file detected.\n\n"
"This is not supported in the current file format. Running \"upgrade -d\" "
"will remove these duplicate positions. However, this will result in loss "
"of data from the original file!")
tree_sequence.dump(args.destination)
def run_upgrade(args):
try:
tree_sequence = tskit.load_legacy(args.source, args.remove_duplicate_positions)
tree_sequence.dump(args.destination)
except tskit.DuplicatePositionsError:
exit(
"Error: Duplicate mutation positions in the source file detected.\n\n"
"This is not supported in the current file format. Running \"upgrade -d\" "
"will remove these duplicate positions. However, this will result in loss "
"of data from the original file!")
def test_msprime_v_0_3_0(self):
path = os.path.join(test_data_dir, "hdf5-formats",
"msprime-0.3.0_v2.0.hdf5")
ts = tskit.load_legacy(path)
self.verify_tree_sequence(ts)
def test_unknown_legacy_version(self):
root = h5py.File(self.temp_file, "w")
root.attrs["format_version"] = (1024, 0) # Arbitrary unknown version
root.close()
with pytest.raises(ValueError):
tskit.load_legacy(self.temp_file)
def test_no_version_number(self):
root = h5py.File(self.temp_file, "w")
root.attrs["x"] = 0
root.close()
with pytest.raises(ValueError):
tskit.load_legacy(self.temp_file)
