def mask_dnode(self, g: Graph) -> Graph:
"""deprecated"""
g2 = Graph(True, True, True, g.get_n_nodes(), g.get_n_links())
for n in g.iter_nodes():
assert g2.add_new_node(
n.type, n.label if n.type == GraphNodeType.CLASS_NODE else
b"DataNode").id == n.id
for e in g.iter_links():
assert g2.add_new_link(e.type, e.label, e.source_id,
e.target_id).id == e.id
return g2
def build(self, g: Graph) -> GraphExplorer:
# TODO: can make it more efficient by giving estimation to graph explorer
g_explorer = GraphExplorer()
for node in g.iter_nodes():
g_explorer.real_add_new_node(GraphNodeHop(0), node.type,
node.label)
for link in g.iter_links():
g_explorer.add_new_link(link.type, link.label, link.source_id,
link.target_id)
self.explore(g_explorer)
return g_explorer
def convert_graph(graph: Graph):
node_index: Dict[int, Node] = {}
for v in graph.iter_nodes():
type = Node.DATA_NODE if v.is_data_node() else Node.CLASS_NODE
node_index[v.id] = Node(v.id, type, v.label)
for l in graph.iter_links():
if data_node_mode == 2:
if node_index[l.target_id].type == Node.DATA_NODE:
# ignore data node
continue
link = Link(l.id, l.label, l.source_id, l.target_id)
Node.add_outgoing_link(node_index[l.source_id], link)
Node.add_incoming_link(node_index[l.target_id], link)
if data_node_mode == DataNodeMode.IGNORE_DATA_NODE:
for v2 in [
v for v in node_index.values() if v.type == Node.DATA_NODE
]:
del node_index[v2.id]
if data_node_mode == DataNodeMode.IGNORE_LABEL_DATA_NODE:
# we convert label of node to DATA_NODE
leaf_source_nodes: Set[Node] = set()
for v in [
v for v in node_index.values() if v.type == Node.DATA_NODE
]:
assert len(v.incoming_links) == 1
link = v.incoming_links[0]
source = node_index[link.source_id]
leaf_source_nodes.add(source)
for node in leaf_source_nodes:
link_label_count = {}
for link in node.outgoing_links:
target = node_index[link.target_id]
if target.type == Node.DATA_NODE:
if link.label not in link_label_count:
link_label_count[link.label] = 0
link_label_count[link.label] += 1
target.label = 'DATA_NODE' + str(
link_label_count[link.label])
return node_index