try to enable flake8 checker.

.flake8

@@ -0,0 +1,4 @@
+[flake8]
+max-line-length = 120
+per-file-ignores =
+ __init__.py:F401

keras2onnx/_graph_cvt.py

@@ -13,7 +13,7 @@
 # limitations under the License.
 # ==============================================================================
 """Helpers to convert variables to constants in TensorFlow 2.0."""
-
+# flake8: noqa
 from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function

keras2onnx/_parser_1x.py

@@ -100,7 +100,9 @@ def on_parsing_keras_layer(graph, node_list, layer, kenode, model, varset, prefi
  for n_, o_ in enumerate(outputs):
  oname = prefix + o_.name
  k2o_logger().debug('\toutput: ' + oname)
- o1 = varset.get_local_variable_or_declare_one(oname, infer_variable_type(o_, varset.target_opset, kenode_output_shapes[n_]))
+ o1 = varset.get_local_variable_or_declare_one(oname,
+ infer_variable_type(o_, varset.target_opset,
+ kenode_output_shapes[n_]))
  operator.add_output(o1)
 
  if hasattr(layer, 'output_mask') and layer.output_mask is not None:
@@ -122,8 +124,6 @@ def on_parsing_keras_layer(graph, node_list, layer, kenode, model, varset, prefi
 
 
 def build_opdict_from_keras(model):
- # type: (keras.Model) -> {}
-
  output_dict = {}
  for l_ in model.layers:
  if hasattr(l_, 'layers'):

keras2onnx/common/data_types.py

@@ -4,4 +4,4 @@
 # license information.
 ###############################################################################
 
-from onnxconverter_common.data_types import *
+from onnxconverter_common.data_types import * # noqa

keras2onnx/common/onnx_ops.py

@@ -4,6 +4,7 @@
 # license information.
 ###############################################################################
 import functools
+import numpy as np
 import onnxconverter_common
 from onnx.mapping import NP_TYPE_TO_TENSOR_TYPE
 from onnxconverter_common.onnx_ops import * # noqa:

keras2onnx/parser.py

@@ -150,7 +150,7 @@ def _check_layer_converter_availability(sub_model):
  else:
  layer_type = type(l_)
  exist = get_converter(layer_type) or \
-  layer_type in [keras.layers.InputLayer, keras.layers.wrappers.TimeDistributed]
+ layer_type in [keras.layers.InputLayer, keras.layers.wrappers.TimeDistributed]
 
  if not exist:
  k2o_logger().info("The layer {} doesn't have a specific converter, fall back.".format(str(l_)))

keras2onnx/proto/tfcompat.py

@@ -36,14 +36,12 @@ def dump_graph_into_tensorboard(tf_graph):
 if is_tf2:
  tensorflow = _tf.compat.v1
 
-
  def is_subclassed(layer):
  """Returns True if the object is a subclassed layer or subclassed model."""
  return (layer.__module__.find('keras.engine') == -1 and
  layer.__module__.find('keras.layers') == -1)
 else:
  tensorflow = _tf
 
-
  def is_subclassed(layer):
  return False

keras2onnx/topology.py

@@ -137,7 +137,8 @@ def _check_structure(self):
  # A operator has an output, so we remove the operator from the unused-operator list.
  unused_operators.discard(operator.full_name)
  for variable in operator.input_masks + operator.output_masks:
- if variable is None: continue
+ if variable is None:
+ continue
  # A variable is used by an operator, so we remove the variable from the unused-variable list.
  unused_variables.discard(variable.full_name)
  # A operator has an output, so we remove the operator from the unused-operator list.
@@ -236,7 +237,8 @@ def _remove_unused_nodes(nodes, inputs, outputs):
 def _build_extra_inputs(container):
  # When calling ModelComponentContainer's add_initializer(...), nothing is added into the input list.
  # However, In ONNX, for target opset < 9, initializers should also be model's (GraphProto) inputs.
- # Thus, we create ValueInfoProto objects from initializers (type: TensorProto) directly and then add them into model's input list.
+ # Thus, we create ValueInfoProto objects from initializers (type: TensorProto) ...
+ # ... directly and then add them into model's input list.
  extra_inputs = [] # ValueInfoProto list of the initializers
  for tensor in container.initializers:
  # Sometimes (especially when creating optional input values such as RNN's initial hidden state), an initializer
@@ -351,7 +353,8 @@ def convert_topology(topology, model_name, doc_string, target_opset, channel_fir
  '{} nchw_inputs does not make effect. Please set nchw_inputs to empty.'.format(onnx_not_imported))
  k2o_logger().warning('{} so the convertor optimizer is not enabled.'.format(onnx_not_imported))
  except Exception as e: # noqa
- # either optimizer issue or converter issue, we just let it go to diagnose the issue from the converted model.
+ # either optimizer issue or converter issue, we just let it go...
+ # ... so that we can diagnose the issue from the converted model.
  k2o_logger().warning(
  'There is an error({}) happened during optimizing on the converted model!'.format(type(e)))
  k2o_logger().warning(str(e))