def get_index_from_counts(counts):
"""Return index generated from counts
This function return the index from given counts.
For example, when counts = [ 2, 3, 4], return [0, 2, 5, 9]
Args:
counts: numpy.ndarray of paddle.Tensor
Return:
Return idnex of the counts
"""
if check_is_tensor(counts):
index = paddle.concat(
[paddle.zeros(shape=[
1,
], dtype="int64"),
paddle.cumsum(counts)],
axis=-1)
else:
index = np.cumsum(counts, dtype="int64")
index = np.insert(index, 0, 0)
return index
def from_index(cls, sorted_v, sorted_u, sorted_eid, degree, indptr):
self = cls()
self._degree = degree
self._sorted_v = sorted_v
self._sorted_u = sorted_u
self._sorted_eid = sorted_eid
self._indptr = indptr
self._is_tensor = check_is_tensor(sorted_v, sorted_u, sorted_eid,
degree, indptr)
return self
def from_edges(cls, u, v, num_nodes):
self = cls()
self._is_tensor = check_is_tensor(u, v, num_nodes)
if self._is_tensor:
self._degree = paddle.zeros(shape=[num_nodes], dtype="int64")
self._degree = scatter(x=self._degree,
overwrite=False,
index=u,
updates=paddle.ones_like(u, dtype="int64"))
self._sorted_eid = paddle.argsort(u)
self._sorted_u = paddle.gather(u, self._sorted_eid)
self._sorted_v = paddle.gather(v, self._sorted_eid)
self._indptr = op.get_index_from_counts(self._degree)
else:
self._degree, self._sorted_v, self._sorted_u, \
self._sorted_eid, self._indptr = graph_kernel.build_index(u, v, num_nodes)
return self
def __init__(self,
edges,
num_nodes=None,
node_feat=None,
edge_feat=None,
**kwargs):
if node_feat is not None:
self._node_feat = node_feat
else:
self._node_feat = {}
if edge_feat is not None:
self._edge_feat = edge_feat
else:
self._edge_feat = {}
if not check_is_tensor(edges):
if isinstance(edges, np.ndarray):
if edges.dtype != "int64":
edges = edges.astype("int64")
else:
edges = np.array(edges, dtype="int64")
self._edges = edges
if num_nodes is None:
self._num_nodes = maybe_num_nodes(self._edges)
else:
self._num_nodes = num_nodes
max_edge_id = maybe_num_nodes(self._edges)
if not isinstance(max_edge_id, paddle.fluid.framework.Variable
) and self._num_nodes < max_edge_id:
raise ValueError("The max edge ID should be less than the number of nodes. "
"But got max edge ID [%s] >= num_nodes [%s]" \
% (max_edge_id-1, self._num_nodes))
self._adj_src_index = kwargs.get("adj_src_index", None)
self._adj_dst_index = kwargs.get("adj_dst_index", None)
if check_is_tensor(self._num_nodes, self._edges,
*list(self._node_feat.values()),
*list(self._edge_feat.values())):
self._is_tensor = True
elif self._adj_src_index is not None and self._adj_src_index.is_tensor(
):
self._is_tensor = True
elif self._adj_dst_index is not None and self._adj_dst_index.is_tensor(
):
self._is_tensor = True
else:
self._is_tensor = False
if self._is_tensor:
# ensure all variable is tenosr
if not check_is_tensor(self._num_nodes):
self._num_nodes = paddle.to_tensor(self._num_nodes)
if not check_is_tensor(self._edges):
self._edges = paddle.to_tensor(self._edges)
for key in self._node_feat:
if not check_is_tensor(self._node_feat[key]):
self._node_feat[key] = paddle.to_tensor(
self._node_feat[key])
for key in self._edge_feat:
if not check_is_tensor(self._edge_feat[key]):
self._edge_feat[key] = paddle.to_tensor(
self._edge_feat[key])
if self._adj_src_index is not None:
if not self._adj_src_index.is_tensor():
self._adj_src_index.tensor(inplace=True)
if self._adj_dst_index is not None:
if not self._adj_dst_index.is_tensor():
self._adj_dst_index.tensor(inplace=True)
# preprocess graph level informations
self._process_graph_info(**kwargs)
self._nodes = None
def __init__(self,
edges,
num_nodes=None,
node_feat=None,
edge_feat=None,
**kwargs):
if node_feat is not None:
self._node_feat = node_feat
else:
self._node_feat = {}
if edge_feat is not None:
self._edge_feat = edge_feat
else:
self._edge_feat = {}
if not check_is_tensor(edges):
if isinstance(edges, np.ndarray):
if edges.dtype != "int64":
edges = edges.astype("int64")
else:
edges = np.array(edges, dtype="int64")
self._edges = edges
if num_nodes is None:
self._num_nodes = maybe_num_nodes(self._edges)
else:
self._num_nodes = num_nodes
self._adj_src_index = kwargs.get("adj_src_index", None)
self._adj_dst_index = kwargs.get("adj_dst_index", None)
if check_is_tensor(self._num_nodes, self._edges,
*list(self._node_feat.values()),
*list(self._edge_feat.values())):
self._is_tensor = True
elif self._adj_src_index is not None and self._adj_src_index.is_tensor(
):
self._is_tensor = True
elif self._adj_dst_index is not None and self._adj_dst_index.is_tensor(
):
self._is_tensor = True
else:
self._is_tensor = False
if self._is_tensor:
# ensure all variable is tenosr
if not check_is_tensor(self._num_nodes):
self._num_nodes = paddle.to_tensor(self._num_nodes)
if not check_is_tensor(self._edges):
self._edges = paddle.to_tensor(self._edges)
for key in self._node_feat:
if not check_is_tensor(self._node_feat[key]):
self._node_feat[key] = paddle.to_tensor(
self._node_feat[key])
for key in self._edge_feat:
if not check_is_tensor(self._edge_feat[key]):
self._edge_feat[key] = paddle.to_tensor(
self._edge_feat[key])
if self._adj_src_index is not None:
if not self._adj_src_index.is_tensor():
self._adj_src_index.tensor(inplace=True)
if self._adj_dst_index is not None:
if not self._adj_dst_index.is_tensor():
self._adj_dst_index.tensor(inplace=True)
# preprocess graph level informations
self._process_graph_info(**kwargs)
self._nodes = None
