Fix None shape error when one input to ConcatV2 has a shape.

I tried to convert a network where one of the inputs to a concatenation along dimension -1 had shape None. The other input did however have a shape. The conversion failed because the code only looked att the shape of the first input to determine what positive axis value to use in the concatenation. If the order of the inputs had been reversed, the conversion would have worked. I have now changed the code to look at the shapes of both input nodes. With the new code, I can convert the network. I have also verified that the resulting onnx-file works.

Author: klasma

tf2onnx/onnx_opset/tensor.py

@@ -297,8 +297,13 @@ def version_1(cls, ctx, node, **kwargs):
  ctx.remove_input(node, node.input[-1], len(node.input) - 1)
 
  if axis_val < 0: # onnxruntime does not support -1 axis, but TF supports.
- input_shape = ctx.get_shape(node.input[0])
- utils.make_sure(input_shape is not None, "shape of {} is None".format(node.input[0]))
+ input_shape = None
+ for node_input in node.input:
+ input_shape = ctx.get_shape(node_input)
+ if input_shape is not None:
+ break
+ utils.make_sure(input_shape is not None,
+ "the shapes of the following inputs are None: {}".format(', '.join(node.input)))
  axis_val = len(input_shape) + axis_val
  node.set_attr("axis", axis_val)