Fix paddle.incubate.graph_reindex divide by 0 error (#51714)

paddle/phi/kernels/cpu/graph_reindex_kernel.cc

@@ -42,6 +42,10 @@ void GraphReindexKernel(const Context& dev_ctx,
   std::unordered_map<T, T> node_map;
   std::vector<T> unique_nodes;
   int reindex_id = 0;
+  PADDLE_ENFORCE_NE(
+      0,
+      bs,
+      errors::InvalidArgument("The first of dims should not be equal to 0."));
   for (int i = 0; i < bs; i++) {
     T node = x_data[i];
     unique_nodes.emplace_back(node);

paddle/phi/kernels/gpu/graph_reindex_kernel.cu

@@ -389,6 +389,10 @@ void GraphReindexKernel(const Context& dev_ctx,
   const T* neighbors_data = neighbors.data<T>();
   const int* count_data = count.data<int>();
   const int bs = x.dims()[0];
+  PADDLE_ENFORCE_NE(
+      0,
+      bs,
+      errors::InvalidArgument("The first of dims should not be equal to 0."));
   const int num_edges = neighbors.dims()[0];
   reindex_src->Resize({num_edges});
 

python/paddle/fluid/tests/unittests/test_graph_reindex.py

@@ -212,6 +212,20 @@ class TestGraphReindex(unittest.TestCase):
             )
             np.testing.assert_allclose(self.out_nodes, out_nodes_2, rtol=1e-05)
 
+    def test_reindex_div_zero(self):
+        paddle.disable_static()
+        array = np.array([], dtype=np.float32)
+        x = paddle.to_tensor(np.reshape(array, [0]), dtype='int32')
+        with self.assertRaises(ValueError):
+            paddle.incubate.graph_reindex(
+                x=x,
+                neighbors=x,
+                count=x,
+                value_buffer=x,
+                index_buffer=x,
+                flag_buffer_hashtable=False,
+            )
+
 
 class TestGeometricGraphReindex(unittest.TestCase):
     def setUp(self):