rewrite code-like comments to avoid CI syntax warnings

Add error info. for input and output node name check

mindinsight/mindconverter/graph_based_converter/mapper/impl/nn/conv_mapper.py

@@ -100,7 +100,7 @@ class ConvMapper(ONNXToMindSporeMapper):
             stride = params.get('strides')
 
         kernel_size = params.get('kernel_shape')
-        # Onnx inchannel = ms inchannel / group
+        # Onnx in_channel equals ms inchannel divided by group
         in_channels = weight.shape[1] * params.get('group', 1)
         out_channels = weight.shape[0]
         if len(kernel_size) == 1:

mindinsight/mindconverter/graph_based_converter/third_party_graph/onnx_utils.py

@@ -52,6 +52,17 @@ def convert_tf_graph_to_onnx(model_path, model_inputs, model_outputs, opset=None
 
     target = ",".join(constants.DEFAULT_TARGET)
     shape_override = None
+    if not 'input' in model_inputs:
+        error_msg = "The given input node is not an eligible input node."
+        error = ValueError(error_msg)
+        log.error(str(error))
+        raise error
+
+    if 'input' in model_outputs:
+        error_msg = "The given output node is an input node."
+        error = ValueError(error_msg)
+        log.error(str(error))
+        raise error
 
     if model_inputs:
         model_inputs, shape_override = utils.split_nodename_and_shape(
@@ -254,8 +265,7 @@ class OnnxDataLoader:
 
         self.nodes_dict = OrderedDict()  # {node_name: OnnxNode} NO INPUT NODE
         self.tensors_dict = {}  # {tensor_name: OnnxTensor}
-        # {node_name : (type, dim)} NO INPUT & OUTPUT NODE
-        self.value_info_dict = {}
+        self.value_info_dict = {}  # Not contains input and output nodes
 
         self.tensor_name_set = set()  # [str]
         self.node_name_set = set()  # [str]
@@ -395,7 +405,7 @@ class OnnxDataLoader:
         for node_name, node in self.nodes_dict.items():
             # for each input of a node
             for input_name in node.input_name_list:
-                # input_name = input_node:0, remove :0 here
+                # remove :0 in the name to ensure consistency in hierarical tree.
                 input_name = input_name.split(':')[0]
                 if input_name in self.node_name_set:
                     # input is a node