[BUGFIX] Fix is_multigraph in the construction from scipy coo matrix (#1357)

* fix is_multigraph in from_coo.

* add tests for partition.

* fix.

* Revert "add tests for partition."

This reverts commit cb8c8555da3e0c70a482c2d639adce2943475bfc.

* fix everywhere from_scipy_sparse_matrix is used.

python/dgl/graph.py

@@ -1805,7 +1805,7 @@ class DGLGraph(DGLBaseGraph):
                         raise DGLError('Not all edges have attribute {}.'.format(attr))
                 self._edge_frame[attr] = _batcher(attr_dict[attr])
 
-    def from_scipy_sparse_matrix(self, spmat):
+    def from_scipy_sparse_matrix(self, spmat, multigraph=False):
         """ Convert from scipy sparse matrix.
 
         Parameters
@@ -1813,6 +1813,10 @@ class DGLGraph(DGLBaseGraph):
         spmat : scipy sparse matrix
             The graph's adjacency matrix
 
+        multigraph : bool, optional
+            Whether the graph would be a multigraph. If the input scipy sparse matrix is CSR,
+            this argument is ignored.
+
         Examples
         --------
         >>> from scipy.sparse import coo_matrix
@@ -1824,7 +1828,7 @@ class DGLGraph(DGLBaseGraph):
         >>> g.from_scipy_sparse_matrix(a)
         """
         self.clear()
-        self._graph = graph_index.from_scipy_sparse_matrix(spmat, self.is_readonly)
+        self._graph = graph_index.from_scipy_sparse_matrix(spmat, multigraph, self.is_readonly)
         self._node_frame.add_rows(self.number_of_nodes())
         self._edge_frame.add_rows(self.number_of_edges())
         self._msg_frame.add_rows(self.number_of_edges())

python/dgl/graph_index.py

@@ -1139,12 +1139,15 @@ def from_networkx(nx_graph, readonly):
     dst = utils.toindex(dst)
     return from_coo(num_nodes, src, dst, is_multigraph, readonly)
 
-def from_scipy_sparse_matrix(adj, readonly):
+def from_scipy_sparse_matrix(adj, multigraph, readonly):
     """Convert from scipy sparse matrix.
 
     Parameters
     ----------
     adj : scipy sparse matrix
+    multigraph : bool
+        Whether the graph would be a multigraph. If none, the flag will be determined
+        by the data.
     readonly : bool
         True if the returned graph is readonly.
 
@@ -1156,8 +1159,9 @@ def from_scipy_sparse_matrix(adj, readonly):
     if adj.getformat() != 'csr' or not readonly:
         num_nodes = max(adj.shape[0], adj.shape[1])
         adj_coo = adj.tocoo()
-        return from_coo(num_nodes, adj_coo.row, adj_coo.col, False, readonly)
+        return from_coo(num_nodes, adj_coo.row, adj_coo.col, multigraph, readonly)
     else:
+        # If the input matrix is csr, it's guaranteed to be a simple graph.
         return from_csr(adj.indptr, adj.indices, False, "out")
 
 def from_edge_list(elist, is_multigraph, readonly):
@@ -1298,7 +1302,7 @@ def create_graph_index(graph_data, multigraph, readonly):
         return from_edge_list(graph_data, multigraph, readonly)
     elif isinstance(graph_data, scipy.sparse.spmatrix):
         # scipy format
-        return from_scipy_sparse_matrix(graph_data, readonly)
+        return from_scipy_sparse_matrix(graph_data, multigraph, readonly)
     else:
         # networkx - any format
         try:

tests/compute/test_transform.py

@@ -202,7 +202,8 @@ def create_large_graph_index(num_nodes):
     row = np.random.choice(num_nodes, num_nodes * 10)
     col = np.random.choice(num_nodes, num_nodes * 10)
     spm = spsp.coo_matrix((np.ones(len(row)), (row, col)))
-    return from_scipy_sparse_matrix(spm, True)
+    # It's possible that we generate a multigraph.
+    return from_scipy_sparse_matrix(spm, True, True)
 
 def get_nodeflow(g, node_ids, num_layers):
     batch_size = len(node_ids)

tests/graph_index/test_subgraph.py

@@ -92,7 +92,8 @@ def create_large_graph_index(num_nodes):
     row = np.random.choice(num_nodes, num_nodes * 10)
     col = np.random.choice(num_nodes, num_nodes * 10)
     spm = spsp.coo_matrix((np.ones(len(row)), (row, col)))
-    return from_scipy_sparse_matrix(spm, True)
+    # It's possible that we generate a multigraph.
+    return from_scipy_sparse_matrix(spm, True, True)
 
 def test_node_subgraph_with_halo():
     gi = create_large_graph_index(1000)