[GraphBolt] Add implementation of num_nodes (#6395)

Co-authored-by: Ubuntu <ubuntu@ip-172-31-39-125.ap-northeast-1.compute.internal>
Co-authored-by: Hongzhi (Steve), Chen <chenhongzhi.nkcs@gmail.com>

python/dgl/graphbolt/impl/csc_sampling_graph.py

@@ -12,6 +12,7 @@ from ...base import ETYPE
 from ...convert import to_homogeneous
 from ...heterograph import DGLGraph
 from ..base import etype_str_to_tuple, etype_tuple_to_str, ORIGINAL_EDGE_ID
+from ..sampling_graph import SamplingGraph
 from .sampled_subgraph_impl import SampledSubgraphImpl
 
 
@@ -74,7 +75,7 @@ class GraphMetadata:
         self.edge_type_to_id = edge_type_to_id
 
 
-class CSCSamplingGraph:
+class CSCSamplingGraph(SamplingGraph):
     r"""Class for CSC sampling graph."""
 
     def __repr__(self):
@@ -83,6 +84,7 @@ class CSCSamplingGraph:
     def __init__(
         self, c_csc_graph: torch.ScriptObject, metadata: Optional[GraphMetadata]
     ):
+        super().__init__()
         self._c_csc_graph = c_csc_graph
         self._metadata = metadata
 
@@ -108,6 +110,54 @@ class CSCSamplingGraph:
         """
         return self._c_csc_graph.num_edges()
 
+    @property
+    def num_nodes(self) -> Union[int, Dict[str, int]]:
+        """The number of nodes in the graph.
+        - If the graph is homogenous, returns an integer.
+        - If the graph is heterogenous, returns a dictionary.
+
+        Returns
+        -------
+        Union[int, Dict[str, int]]
+            The number of nodes. Integer indicates the total nodes number of a
+            homogenous graph; dict indicates nodes number per node types of a
+            heterogenous graph.
+
+        Examples
+        --------
+        >>> import dgl.graphbolt as gb, torch
+        >>> total_num_nodes = 5
+        >>> total_num_edges = 12
+        >>> ntypes = {"N0": 0, "N1": 1}
+        >>> etypes = {"N0:R0:N0": 0, "N0:R1:N1": 1,
+        ... "N1:R2:N0": 2, "N1:R3:N1": 3}
+        >>> indptr = torch.LongTensor([0, 3, 5, 7, 9, 12])
+        >>> indices = torch.LongTensor([0, 1, 4, 2, 3, 0, 1, 1, 2, 0, 3, 4])
+        >>> node_type_offset = torch.LongTensor([0, 2, 5])
+        >>> type_per_edge = torch.LongTensor(
+        ... [0, 0, 2, 2, 2, 1, 1, 1, 3, 1, 3, 3])
+        >>> metadata = gb.GraphMetadata(ntypes, etypes)
+        >>> graph = gb.from_csc(indptr, indices, node_type_offset,
+        ... type_per_edge, None, metadata)
+        >>> print(graph.num_nodes)
+        {'N0': tensor(2), 'N1': tensor(3)}
+        """
+
+        offset = self.node_type_offset
+
+        # Homogenous.
+        if offset is None or self.metadata is None:
+            return self._c_csc_graph.num_nodes()
+
+        # Heterogenous
+        else:
+            num_nodes_per_type = {
+                _type: offset[_idx + 1] - offset[_idx]
+                for _type, _idx in self.metadata.node_type_to_id.items()
+            }
+
+            return num_nodes_per_type
+
     @property
     def csc_indptr(self) -> torch.tensor:
         """Returns the indices pointer in the CSC graph.
@@ -312,8 +362,8 @@ class CSCSamplingGraph:
             without replacement. If True, a value can be selected multiple
             times. Otherwise, each value can be selected only once.
         probs_name: str, optional
-            An optional string specifying the name of an edge attribute used a. This
-            attribute tensor should contain (unnormalized) probabilities
+            An optional string specifying the name of an edge attribute used.
+            This attribute tensor should contain (unnormalized) probabilities
             corresponding to each neighboring edge of a node. It must be a 1D
             floating-point or boolean tensor, with the number of elements
             equalling the total number of edges.

python/dgl/graphbolt/sampling_graph.py

@@ -23,18 +23,3 @@ class SamplingGraph:
             heterogenous graph.
         """
         raise NotImplementedError
-
-    @property
-    def num_edges(self) -> Union[int, Dict[str, int]]:
-        """The number of edges in the graph.
-        - If the graph is homogenous, returns an integer.
-        - If the graph is heterogenous, returns a dictionary.
-
-        Returns
-        -------
-        Union[int, Dict[str, int]]
-            The number of edges. Integer indicates the total edges number of a
-            homogenous graph; dict indicates edges number per edge types of a
-            heterogenous graph.
-        """
-        raise NotImplementedError

tests/python/pytorch/graphbolt/impl/test_csc_sampling_graph.py

@@ -177,6 +177,84 @@ def test_hetero_graph(total_num_nodes, total_num_edges, num_ntypes, num_etypes):
     assert metadata.edge_type_to_id == graph.metadata.edge_type_to_id
 
 
+@unittest.skipIf(
+    F._default_context_str == "gpu",
+    reason="Graph is CPU only at present.",
+)
+@pytest.mark.parametrize(
+    "total_num_nodes, total_num_edges",
+    [(1, 1), (100, 1), (10, 50), (1000, 50000)],
+)
+def test_num_nodes_homo(total_num_nodes, total_num_edges):
+    csc_indptr, indices = gbt.random_homo_graph(
+        total_num_nodes, total_num_edges
+    )
+    edge_attributes = {
+        "A1": torch.randn(total_num_edges),
+        "A2": torch.randn(total_num_edges),
+    }
+    graph = gb.from_csc(csc_indptr, indices, edge_attributes=edge_attributes)
+
+    assert graph.num_nodes == total_num_nodes
+
+
+@unittest.skipIf(
+    F._default_context_str == "gpu",
+    reason="Graph is CPU only at present.",
+)
+def test_num_nodes_hetero():
+    """Original graph in COO:
+    1   0   1   0   1
+    1   0   1   1   0
+    0   1   0   1   0
+    0   1   0   0   1
+    1   0   0   0   1
+
+    node_type_0: [0, 1]
+    node_type_1: [2, 3, 4]
+    edge_type_0: node_type_0 -> node_type_0
+    edge_type_1: node_type_0 -> node_type_1
+    edge_type_2: node_type_1 -> node_type_0
+    edge_type_3: node_type_1 -> node_type_1
+    """
+    # Initialize data.
+    total_num_nodes = 5
+    total_num_edges = 12
+    ntypes = {
+        "N0": 0,
+        "N1": 1,
+    }
+    etypes = {
+        "N0:R0:N0": 0,
+        "N0:R1:N1": 1,
+        "N1:R2:N0": 2,
+        "N1:R3:N1": 3,
+    }
+    indptr = torch.LongTensor([0, 3, 5, 7, 9, 12])
+    indices = torch.LongTensor([0, 1, 4, 2, 3, 0, 1, 1, 2, 0, 3, 4])
+    node_type_offset = torch.LongTensor([0, 2, 5])
+    type_per_edge = torch.LongTensor([0, 0, 2, 2, 2, 1, 1, 1, 3, 1, 3, 3])
+    assert indptr[-1] == total_num_edges
+    assert indptr[-1] == len(indices)
+    assert node_type_offset[-1] == total_num_nodes
+    assert all(type_per_edge < len(etypes))
+
+    # Construct CSCSamplingGraph.
+    metadata = gb.GraphMetadata(ntypes, etypes)
+    graph = gb.from_csc(
+        indptr, indices, node_type_offset, type_per_edge, None, metadata
+    )
+
+    # Verify nodes number per node types.
+    assert graph.num_nodes == {
+        "N0": 2,
+        "N1": 3,
+    }
+    assert graph.num_nodes["N0"] == 2
+    assert graph.num_nodes["N1"] == 3
+    assert "N2" not in graph.num_nodes
+
+
 @unittest.skipIf(
     F._default_context_str == "gpu",
     reason="Graph is CPU only at present.",