def test_extract_nodegraph_info_badfile():
try:
khmer.extract_nodegraph_info(
utils.get_test_data('test-abund-read-2.fa'))
assert 0, 'this should fail'
except ValueError:
pass
def test_extract_nodegraph_info_badfile():
try:
khmer.extract_nodegraph_info(
utils.get_test_data('test-abund-read-2.fa'))
assert 0, 'this should fail'
except ValueError:
pass
def _load_v2(cls,
info,
leaf_loader,
dirname,
storage,
print_version_warning=True):
nodes = {int(k): v for (k, v) in info['nodes'].items()}
if nodes[0] is None:
raise ValueError("Empty tree!")
sbt_nodes = defaultdict(lambda: None)
sample_bf = os.path.join(dirname, nodes[0]['filename'])
k, size, ntables = khmer.extract_nodegraph_info(sample_bf)[:3]
factory = GraphFactory(k, size, ntables)
for k, node in nodes.items():
if node is None:
continue
node['filename'] = os.path.join(dirname, node['filename'])
if 'internal' in node['name']:
node['factory'] = factory
sbt_node = Node.load(node, storage)
else:
sbt_node = leaf_loader(node, storage)
sbt_nodes[k] = sbt_node
tree = cls(factory, d=info['d'])
tree.nodes = sbt_nodes
return tree
def _load_v2(cls, info, leaf_loader, dirname, storage, print_version_warning=True):
nodes = {int(k): v for (k, v) in info['nodes'].items()}
if nodes[0] is None:
raise ValueError("Empty tree!")
sbt_nodes = defaultdict(lambda: None)
sample_bf = os.path.join(dirname, nodes[0]['filename'])
k, size, ntables = khmer.extract_nodegraph_info(sample_bf)[:3]
factory = GraphFactory(k, size, ntables)
for k, node in nodes.items():
if node is None:
continue
node['filename'] = os.path.join(dirname, node['filename'])
if 'internal' in node['name']:
node['factory'] = factory
sbt_node = Node.load(node, storage)
else:
sbt_node = leaf_loader(node, storage)
sbt_nodes[k] = sbt_node
tree = cls(factory, d=info['d'])
tree.nodes = sbt_nodes
return tree
def _load_v1(jnodes,
leaf_loader,
dirname,
storage,
print_version_warning=True):
if jnodes[0] is None:
raise ValueError("Empty tree!")
sbt_nodes = defaultdict(lambda: None)
sample_bf = os.path.join(dirname, jnodes[0]['filename'])
ksize, tablesize, ntables = khmer.extract_nodegraph_info(sample_bf)[:3]
factory = GraphFactory(ksize, tablesize, ntables)
for i, jnode in enumerate(jnodes):
if jnode is None:
continue
jnode['filename'] = os.path.join(dirname, jnode['filename'])
if 'internal' in jnode['name']:
jnode['factory'] = factory
sbt_node = Node.load(jnode, storage)
else:
sbt_node = leaf_loader(jnode, storage)
sbt_nodes[i] = sbt_node
tree = SBT(factory)
tree.nodes = sbt_nodes
return tree
def _load_v2(cls, info, leaf_loader, dirname):
nodes = {int(k): v for (k, v) in info['nodes'].items()}
if nodes[0] is None:
raise ValueError("Empty tree!")
sbt_nodes = []
sample_bf = os.path.join(dirname, nodes[0]['filename'])
k, size, ntables = khmer.extract_nodegraph_info(sample_bf)[:3]
factory = GraphFactory(k, size, ntables)
for i, node in sorted(nodes.items()):
if node is None:
sbt_nodes.append(None)
continue
if 'internal' in node['filename']:
node['factory'] = factory
sbt_node = LazyNode(Node.load, node, dirname)
else:
sbt_node = LazyNode(leaf_loader, node, dirname)
sbt_nodes.append(sbt_node)
tree = cls(factory, d=info['d'])
tree.nodes = sbt_nodes
return tree
def _load_v1(jnodes, leaf_loader, dirname, storage, print_version_warning=True):
if jnodes[0] is None:
raise ValueError("Empty tree!")
sbt_nodes = defaultdict(lambda: None)
sample_bf = os.path.join(dirname, jnodes[0]['filename'])
ksize, tablesize, ntables = khmer.extract_nodegraph_info(sample_bf)[:3]
factory = GraphFactory(ksize, tablesize, ntables)
for i, jnode in enumerate(jnodes):
if jnode is None:
continue
jnode['filename'] = os.path.join(dirname, jnode['filename'])
if 'internal' in jnode['name']:
jnode['factory'] = factory
sbt_node = Node.load(jnode, storage)
else:
sbt_node = leaf_loader(jnode, storage)
sbt_nodes[i] = sbt_node
tree = SBT(factory)
tree.nodes = sbt_nodes
return tree
def _load_v1(jnodes, leaf_loader, dirname):
if jnodes[0] is None:
# TODO error!
raise ValueError("Empty tree!")
sbt_nodes = []
sample_bf = os.path.join(dirname, jnodes[0]['filename'])
ksize, tablesize, ntables = khmer.extract_nodegraph_info(sample_bf)[:3]
factory = GraphFactory(ksize, tablesize, ntables)
for jnode in jnodes:
if jnode is None:
sbt_nodes.append(None)
continue
if 'internal' in jnode['filename']:
jnode['factory'] = factory
sbt_node = LazyNode(Node.load, jnode, dirname)
else:
sbt_node = LazyNode(leaf_loader, jnode, dirname)
sbt_nodes.append(sbt_node)
tree = SBT(factory)
tree.nodes = sbt_nodes
return tree
def load(sbt_fn):
with open(sbt_fn) as fp:
nodes = json.load(fp)
if nodes[0] is None:
# TODO error!
raise ValueError("Empty tree!")
sbt_nodes = []
ksize, tablesize, ntables, _, _, _ = khmer.extract_nodegraph_info(nodes[0]["filename"])
factory = GraphFactory(ksize, tablesize, ntables)
for node in nodes:
if node is None:
sbt_nodes.append(None)
continue
graph = khmer.load_nodegraph(node["filename"])
if "metadata" in node:
# only Leaf nodes have metadata
l = Leaf(node["metadata"], graph)
sbt_nodes.append(l)
else:
n = Node(factory, name=node["name"])
n.graph = graph
sbt_nodes.append(n)
tree = SBT(factory)
tree.nodes = sbt_nodes
return tree
def load_sbt(sbt_fn):
with open(sbt_fn) as fp:
sbt_dict = json.load(fp)
ksize, tablesize, ntables, _, _, _ = khmer.extract_nodegraph_info(sbt_dict['root']['filename'])
factory = GraphFactory(ksize, tablesize, ntables)
tree = load_node(sbt_dict['root'], factory)
return tree
def test_extract_nodegraph_info():
fn = utils.get_temp_filename("test_extract_nodegraph.pt")
for size in [1e6, 2e6, 5e6, 1e7]:
ht = khmer.Nodegraph(25, size, 4)
ht.save(fn)
info = khmer.extract_nodegraph_info(fn)
ksize, table_size, n_tables, _, _, _ = info
print(ksize, table_size, n_tables)
assert (ksize) == 25
assert table_size == size, table_size
assert n_tables == 4
try:
os.remove(fn)
except OSError as err:
print("...failed to remove {fn}".format(fn) + str(err), file=sys.stderr)
def test_extract_nodegraph_info():
fn = utils.get_temp_filename('test_extract_nodegraph.pt')
for size in [1e6, 2e6, 5e6, 1e7]:
ht = khmer.Nodegraph(25, size, 4)
ht.save(fn)
info = khmer.extract_nodegraph_info(fn)
ksize, table_size, n_tables, _, _, _ = info
print(ksize, table_size, n_tables)
assert(ksize) == 25
assert table_size == size, table_size
assert n_tables == 4
try:
os.remove(fn)
except OSError as err:
print('...failed to remove {fn}'.format(fn) + str(err),
file=sys.stderr)
def load(cls, sbt_fn, leaf_loader=None):
if leaf_loader is None:
leaf_loader = Leaf.load
with open(sbt_fn) as fp:
info = json.load(fp)
nodes = {int(k): v for (k, v) in info['nodes'].items()}
if nodes[0] is None:
raise ValueError("Empty tree!")
sbt_nodes = []
# TODO if there is only one Leaf (the root),
# how do we initialize the factory?
# Probably better to save this in the json
k, size, ntables, *_ = extract_nodegraph_info(nodes[0]['filename'])
factory = GraphFactory(k, size, ntables)
for i, node in sorted(nodes.items()):
if node is None:
sbt_nodes.append(None)
continue
if 'internal' in node['filename']:
node['factory'] = factory
new_node = Node.load(node)
else:
new_node = leaf_loader(node)
sbt_nodes.append(new_node)
tree = cls(factory, d=info['d'])
tree.nodes = sbt_nodes
return tree
