upgrade to 1.7

onnxmltools/__init__.py

@@ -9,7 +9,7 @@
 This framework converts any machine learned model into onnx format
 which is a common language to describe any machine learned model.
 """
-__version__ = "1.6.5"
+__version__ = "1.7.0"
 __author__ = "Microsoft"
 __producer__ = "OnnxMLTools"
 __producer_version__ = __version__

onnxmltools/convert/coreml/convert.py

@@ -6,8 +6,9 @@
 
 import coremltools
 from uuid import uuid4
+from onnxconverter_common.onnx_ex import get_maximum_opset_supported
 from ...proto import onnx
-from ...proto import onnx_proto, get_opset_number_from_onnx
+from ...proto import onnx_proto
 from ..common._topology import convert_topology
 from ._parse import parse_coreml
 
@@ -55,7 +56,7 @@ def convert(model, name=None, initial_types=None, doc_string='', target_opset=No
     if name is None:
         name = str(uuid4().hex)
 
-    target_opset = target_opset if target_opset else get_opset_number_from_onnx()
+    target_opset = target_opset if target_opset else get_maximum_opset_supported()
     # Parse CoreML model as our internal data structure (i.e., Topology)
     topology = parse_coreml(spec, initial_types, target_opset, custom_conversion_functions, custom_shape_calculators)
 

onnxmltools/convert/h2o/convert.py

@@ -9,7 +9,8 @@
 import tempfile
 import h2o
 
-from ...proto import onnx, get_opset_number_from_onnx
+from onnxconverter_common.onnx_ex import get_maximum_opset_supported
+from ...proto import onnx
 from ..common._topology import convert_topology
 from ..common.data_types import FloatTensorType
 from ._parse import parse_h2o
@@ -64,7 +65,7 @@ def convert(model, name=None, initial_types=None, doc_string='', target_opset=No
     if mojo_model["params"]["algo"] != "gbm":
         raise ValueError("Model type not supported (algo=%s). Only GBM Mojo supported for now." % mojo_model["params"]["algo"])
 
-    target_opset = target_opset if target_opset else get_opset_number_from_onnx()
+    target_opset = target_opset if target_opset else get_maximum_opset_supported()
     topology = parse_h2o(mojo_model, initial_types, target_opset, custom_conversion_functions, custom_shape_calculators)
     topology.compile()
     onnx_model = convert_topology(topology, name, doc_string, target_opset, targeted_onnx)

onnxmltools/convert/libsvm/convert.py

@@ -5,6 +5,7 @@
 #--------------------------------------------------------------------------
 
 from uuid import uuid4
+from onnxconverter_common.onnx_ex import get_maximum_opset_supported
 from ...proto import onnx
 from ..common._topology import convert_topology
 from ._parse import parse_libsvm
@@ -34,6 +35,7 @@ def convert(model, name=None, initial_types=None, doc_string='', target_opset=No
 
     if name is None:
         name = str(uuid4().hex)
+    target_opset = target_opset if target_opset else get_maximum_opset_supported()
 
     # Parse scikit-learn model as our internal data structure (i.e., Topology)
     topology = parse_libsvm(model, initial_types, custom_conversion_functions, 

onnxmltools/convert/lightgbm/convert.py

@@ -6,7 +6,8 @@
 
 from uuid import uuid4
 import lightgbm
-from ...proto import onnx, get_opset_number_from_onnx
+from onnxconverter_common.onnx_ex import get_maximum_opset_supported
+from ...proto import onnx
 from ..common._topology import convert_topology
 from ._parse import parse_lightgbm, WrappedBooster
 
@@ -45,7 +46,7 @@ def convert(model, name=None, initial_types=None, doc_string='', target_opset=No
     if name is None:
         name = str(uuid4().hex)
 
-    target_opset = target_opset if target_opset else get_opset_number_from_onnx()
+    target_opset = target_opset if target_opset else get_maximum_opset_supported()
     topology = parse_lightgbm(model, initial_types, target_opset, custom_conversion_functions, custom_shape_calculators)
     topology.compile()
     onnx_model = convert_topology(topology, name, doc_string, target_opset, targeted_onnx)

onnxmltools/convert/sparkml/convert.py

@@ -5,7 +5,8 @@
 # --------------------------------------------------------------------------
 
 from uuid import uuid4
-from ...proto import onnx, get_opset_number_from_onnx
+from onnxconverter_common.onnx_ex import get_maximum_opset_supported
+from ...proto import onnx
 from ..common._topology import convert_topology
 from ._parse import parse_sparkml
 from . import operator_converters
@@ -63,7 +64,7 @@ def convert(model, name=None, initial_types=None, doc_string='', target_opset=No
     if name is None:
         name = str(uuid4().hex)
 
-    target_opset = target_opset if target_opset else get_opset_number_from_onnx()
+    target_opset = target_opset if target_opset else get_maximum_opset_supported()
     # Parse spark-ml model as our internal data structure (i.e., Topology)
     topology = parse_sparkml(spark_session, model, initial_types, target_opset, custom_conversion_functions, custom_shape_calculators)
 

onnxmltools/convert/xgboost/convert.py

@@ -6,7 +6,8 @@
 
 from uuid import uuid4
 import xgboost
-from ...proto import onnx, get_opset_number_from_onnx
+from onnxconverter_common.onnx_ex import get_maximum_opset_supported
+from ...proto import onnx
 from ..common._topology import convert_topology
 from ._parse import parse_xgboost, WrappedBooster
 
@@ -40,7 +41,7 @@ def convert(model, name=None, initial_types=None, doc_string='', target_opset=No
 
     if isinstance(model, xgboost.Booster):
         model = WrappedBooster(model)
-    target_opset = target_opset if target_opset else get_opset_number_from_onnx()
+    target_opset = target_opset if target_opset else get_maximum_opset_supported()
     topology = parse_xgboost(model, initial_types, target_opset, custom_conversion_functions, custom_shape_calculators)
     topology.compile()
     onnx_model = convert_topology(topology, name, doc_string, target_opset, targeted_onnx)

onnxmltools/proto/__init__.py

@@ -52,16 +52,3 @@ def _make_tensor_fixed(name, data_type, dims, vals, raw=False):
 
 
 helper.make_tensor = _make_tensor_fixed
-
-
-def get_opset_number_from_onnx():
-    # since the method was widely used among while it is buggy to get the opset number...
-    # ... blindly, so change it to be safer without the name change.
-
-    default_max_opset = 11
-    try:
-        from onnxconverter_common.topology import DEFAULT_OPSET_NUMBER
-        default_max_opset = DEFAULT_OPSET_NUMBER
-    except:  # noqa
-        pass
-    return min(default_max_opset, onnx.defs.onnx_opset_version())

requirements.txt

@@ -1,7 +1,7 @@
 keras2onnx
 numpy
 onnx
-onnxconverter-common>=1.6.5, <1.7.0
+onnxconverter-common>=1.7.0, <1.8.0
 protobuf
 six
 skl2onnx

tests/baseline/test_convert_baseline.py

@@ -50,3 +50,7 @@ def test_keras2coreml_Dense_ImageNet_small(self):
         """
         self.assertFalse(self.check_baseline(
             "keras2coreml_Dense_ImageNet_small.mlmodel", "keras2coreml_Dense_ImageNet_small.json"))
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/xgboost/test_xgboost_pipeline.py

@@ -19,7 +19,8 @@
     from onnxmltools.convert.common.data_types import FloatTensorType
     from onnxmltools.utils import dump_data_and_model
     from onnxmltools.convert.xgboost.operator_converters.XGBoost import convert_xgboost as convert_xgb
-    from onnxmltools.proto import get_opset_number_from_onnx
+    from onnxconverter_common.onnx_ex import get_maximum_opset_supported
+
 
     can_test = True
 except ImportError:
@@ -135,7 +136,7 @@ def common_test_xgboost_10_skl(self, missing, replace=False):
             input_xgb[input_xgb[:, :] == missing] = np.nan
         onnx_last = convert_sklearn(model.steps[1][-1],
                                     initial_types=[('X', FloatTensorType(shape=[None, input_xgb.shape[1]]))],
-                                    target_opset=get_opset_number_from_onnx())
+                                    target_opset=get_maximum_opset_supported())
         session = rt.InferenceSession(onnx_last.SerializeToString())
         pred_skl = model.steps[1][-1].predict(input_xgb).ravel()
         pred_onx = session.run(None, {'X': input_xgb})[0].ravel()