Fix the print issue with the node which has not abstract.

mindspore/ccsrc/frontend/optimizer/irpass/convert.h

@@ -60,7 +60,9 @@ class PrintTupleWrapper : public AnfVisitor {
 
 class PrintConstStringWrapper : public AnfVisitor {
   bool CheckNeedConvert(const AbstractBasePtr &abs) const {
-    MS_EXCEPTION_IF_NULL(abs);
+    if (abs == nullptr) {
+      return false;
+    }
     if (abs->isa<abstract::AbstractSequence>()) {
       auto sequence_abs = abs->cast<abstract::AbstractSequencePtr>();
       const auto &elements = sequence_abs->elements();

mindspore/ccsrc/frontend/optimizer/irpass/inline.h

@@ -305,7 +305,9 @@ class InlinerBase : public AnfVisitor {
     node_inputs.push_back(NewValueNode(new_fg));
     std::transform(used_param_index.begin(), used_param_index.end(), std::back_inserter(node_inputs),
                    [&args](size_t i) { return args[i]; });
-    return node->func_graph()->NewCNode(node_inputs);
+    auto ret_node = node->func_graph()->NewCNode(node_inputs);
+    ret_node->set_abstract(node->abstract());
+    return ret_node;
   }
 
   bool CheckSwitchInputs(const std::vector<AnfNodePtr> &sw_inputs) {
@@ -366,7 +368,6 @@ class InlinerBase : public AnfVisitor {
     return has_branch;
   }
 
- private:
   bool is_checked_{false};
   bool is_recursive_{false};
   // If the user guarantee that fg has no recursive.

tests/st/ops/gpu/test_print_op.py

@@ -17,7 +17,7 @@
 import numpy as np
 import pytest
 
-from mindspore import Tensor
+from mindspore import Tensor, jit
 import mindspore.nn as nn
 from mindspore.ops import operations as P
 import mindspore.context as context
@@ -224,3 +224,32 @@ def test_print_tensor_dtype_in_nested_tuple(mode):
     y = Tensor([1, 2], dtype=ms.int32)
     net = PrintDtypeNet()
     net(x, y)
+
+
+@pytest.mark.level0
+@pytest.mark.env_onecard
+@pytest.mark.platform_x86_gpu_training
+def test_print_abs():
+    """
+    Feature: Print op.
+    Description: Print the result of max.
+    Expectation: success.
+    """
+    @jit
+    def function():
+        tuple_x = (Tensor(10).astype("float32"), Tensor(30).astype("float32"), Tensor(50).astype("float32"))
+        sum_x = Tensor(0).astype("float32")
+        max_x = Tensor(0).astype("float32")
+        for i in range(3):
+            max_x = max(tuple_x)
+            sum_x += max_x
+            print(max_x)
+            print(i)
+        for x in zip(tuple_x):
+            sum_x = sum(x, sum_x)
+            print(sum_x)
+        return sum_x
+
+    context.set_context(mode=context.GRAPH_MODE, device_target="GPU")
+    out = function()
+    print("out:", out)