Avoid direct access to Node::def() where some other method works.

PiperOrigin-RevId: 163704839

tensorflow/core/common_runtime/function.cc

@@ -1030,7 +1030,9 @@ void ToGraphDef(const Graph* g, GraphDef* gdef, bool pretty) {
     NodeDef* ndef = gdef->add_node();
     ndef->set_name(NewName(n, pretty));
     ndef->set_op(n->type_string());
-    *(ndef->mutable_attr()) = n->def().attr();
+    for (const auto& attr : n->attrs()) {
+      (*ndef->mutable_attr())[attr.first] = attr.second;
+    }
     inputs.clear();
     inputs.resize(n->num_inputs());
     for (const Edge* e : n->in_edges()) {

tensorflow/core/common_runtime/simple_placer.cc

@@ -147,7 +147,7 @@ class ColocationGraph {
       // attribute with the calls to ColocateNodeToGroup.
       bool found_spec = false;
       const AttrValue* attr_value =
-          AttrSlice(node->def()).Find(kColocationAttrNameStringPiece);
+          node->attrs().Find(kColocationAttrNameStringPiece);
       if (attr_value != nullptr && attr_value->has_list()) {
         for (const string& class_spec : attr_value->list().s()) {
           StringPiece spec(class_spec);
@@ -184,7 +184,7 @@ class ColocationGraph {
       // error, return it.
       Status s = ColocateNodes(*node, *root_node);
       if (!s.ok()) {
-        return AttachDef(s, node->def());
+        return AttachDef(s, *node);
       }
     }
     return Status::OK();
@@ -418,7 +418,7 @@ class ColocationGraph {
       }
       Status status = InitializeMember(*node, &members_[node->id()]);
       if (!status.ok()) {
-        return AttachDef(status, node->def());
+        return AttachDef(status, *node);
       }
     }
     return Status::OK();
@@ -727,7 +727,7 @@ Status SimplePlacer::Run() {
                                     "be on the same device), but the two nodes "
                                     "were assigned two different devices: ",
                                     status.error_message()),
-            dst->def());
+            *dst);
       }
     }
   }

tensorflow/core/graph/gradients.cc

@@ -110,7 +110,9 @@ static Node* AddSymGrad(Graph* g, Node* n, gtl::ArraySlice<NodeOut> grads) {
   AddNodeAttr("Tout", n->input_types(), &ndef);
   NameAttrList func;
   func.set_name(n->type_string());
-  *(func.mutable_attr()) = n->def().attr();
+  for (const auto& attr : n->attrs()) {
+    (*func.mutable_attr())[attr.first] = attr.second;
+  }
   AddNodeAttr("f", func, &ndef);
   Status s;
   Node* ret = g->AddNode(ndef, &s);

tensorflow/core/graph/optimizer_cse.cc

@@ -189,8 +189,9 @@ bool OptimizerCSE::Optimize(
     if (!n->IsOp()) continue;
 
     // Don't prune placeholder nodes.
-    if (n->def().op() == "Placeholder" || n->def().op() == "PlaceholderV2" ||
+    if (n->type_string() == "Placeholder" ||
-        n->def().op() == "PlaceholderWithDefault") {
+        n->type_string() == "PlaceholderV2" ||
+        n->type_string() == "PlaceholderWithDefault") {
       continue;
     }
 

tensorflow/core/kernels/hexagon/graph_transferer.cc

@@ -376,7 +376,7 @@ Status GraphTransferer::TransformGraphToAddAggregatedInputNode(
     std::vector<DataType> data_types;
     std::vector<TensorShape> shapes;
     Status status = RemoteFusedGraphExecuteUtils::GetOutputTensorShapeType(
-        original_input_node->def(), &data_types, &shapes);
+        original_input_node->attrs(), &data_types, &shapes);
     if (status.ok()) {
       created_node->AddAttr(
           RemoteFusedGraphExecuteUtils::ATTR_OUTPUT_DATA_TYPES, data_types);
@@ -579,7 +579,7 @@ bool GraphTransferer::HasPaddingAndStrides(const Node& node) {
 }
 
 bool GraphTransferer::NeedsToAddRank(const Node& node) {
-  const string& op_type = node.def().op();
+  const StringPiece op_type(node.type_string());
   if (op_type == "Transpose" || op_type == "ExpandDims") {
     return true;
   }
@@ -587,7 +587,7 @@ bool GraphTransferer::NeedsToAddRank(const Node& node) {
 }
 
 bool GraphTransferer::IsPadNode(const Node& node) {
-  const string& op_type = node.def().op();
+  const StringPiece op_type(node.type_string());
   if (op_type == "Pad") {
     return true;
   }
@@ -678,7 +678,7 @@ void GraphTransferer::RegisterNodeWithRank(
   CHECK_NOTNULL(input0_node);
   std::vector<TensorShape> shapes;
   Status status = RemoteFusedGraphExecuteUtils::GetOutputTensorShapeType(
-      input0_node->def(), nullptr, &shapes);
+      input0_node->attrs(), nullptr, &shapes);
   CHECK_EQ(1, shapes.size()) << "Output size should be 1.";
   const int const_val_id =
       RegisterConstScalar(DT_INT32, shapes.at(0).dims(), id, node.num_inputs());
@@ -728,7 +728,7 @@ void GraphTransferer::RegisterPadNode(
   CHECK(input_node->IsConstant());
 
   const TensorProto* tensor_proto = nullptr;
-  TF_CHECK_OK(GetNodeAttr(input_node->def(), "value", &tensor_proto));
+  TF_CHECK_OK(GetNodeAttr(input_node->attrs(), "value", &tensor_proto));
   CHECK_NOTNULL(tensor_proto);
   Tensor const_tensor;
   TF_CHECK_OK(MakeTensorFromProto(*tensor_proto, &const_tensor));
@@ -739,7 +739,7 @@ void GraphTransferer::RegisterPadNode(
   } else if (const_tensor.shape().dim_size(0) < PAD_WIDTH) {
     const int width = const_tensor.shape().dim_size(0);
     const TensorProto* proto = nullptr;
-    TF_CHECK_OK(GetNodeAttr(input_node->def(), "value", &proto));
+    TF_CHECK_OK(GetNodeAttr(input_node->attrs(), "value", &proto));
     Tensor const_tensor;
     TF_CHECK_OK(MakeTensorFromProto(*proto, &const_tensor));
     CHECK_EQ(DT_INT32, const_tensor.dtype());