# -*- encoding: utf-8 -*-
# pylint: disable=E0203,E1101,C0111
"""
Runtime operator.
:githublink:`%|py|7`
"""
from collections import OrderedDict
import numpy
from onnx.defs import onnx_opset_version
from ._op import OpRunUnaryNum
[docs]class Clip_6(OpRunUnaryNum):
atts = {'min': -3.4028234663852886e+38,
'max': 3.4028234663852886e+38}
[docs] def __init__(self, onnx_node, desc=None, **options):
OpRunUnaryNum.__init__(self, onnx_node, desc=desc,
expected_attributes=Clip_6.atts,
**options)
[docs] def _run(self, data): # pylint: disable=W0221
if self.inplaces.get(0, False):
return self._run_inplace(data)
res = numpy.clip(data, self.min, self.max)
return (res, ) if res.dtype == data.dtype else (res.astype(data.dtype), )
[docs] def _run_inplace(self, data):
return (numpy.clip(data, self.min, self.max, out=data), )
[docs] def to_python(self, inputs):
return ("import numpy",
"return numpy.clip(%s, min_, max_)" % inputs[0])
[docs]class Clip_11(OpRunUnaryNum):
version_higher_than = 11
mandatory_inputs = ['X']
optional_inputs = OrderedDict([
('min', -3.4028234663852886e+38),
('max', 3.4028234663852886e+38)
])
[docs] def __init__(self, onnx_node, desc=None, **options):
OpRunUnaryNum.__init__(self, onnx_node, desc=desc,
**options)
[docs] def run(self, x, *minmax): # pylint: disable=E0202,W0221
"""
Calls method ``_run``.
:githublink:`%|py|53`
"""
try:
res = self._run(x, *minmax)
except TypeError as e:
raise TypeError("Issues with types {} (binary operator {}).".format(
", ".join(str(type(_)) for _ in [x]),
self.__class__.__name__)) from e
return res
[docs] def _run(self, data, *minmax): # pylint: disable=W0221
if self.inplaces.get(0, False):
return self._run_inplace(data, *minmax)
le = len(minmax)
amin = minmax[0] if le > 0 else None # -3.4028234663852886e+38
amax = minmax[1] if le > 1 else None # 3.4028234663852886e+38
res = numpy.clip(data, amin, amax)
return (res, ) if res.dtype == data.dtype else (res.astype(data.dtype), )
[docs] def _run_inplace(self, data, *minmax): # pylint: disable=W0221
le = len(minmax)
amin = minmax[0] if le > 0 else None # -3.4028234663852886e+38
amax = minmax[1] if le > 1 else None # 3.4028234663852886e+38
res = numpy.clip(data, amin, amax, out=data)
return (res, )
[docs] def infer_shapes(self, x, *minmax): # pylint: disable=E0202,W0221
try:
return self._infer_shapes(x)
except TypeError as e:
raise TypeError("Issues with types {} (operator {}).".format(
x.dtype, self.__class__.__name__)) from e
[docs] def to_python(self, inputs):
return ("import numpy",
"return numpy.clip(%s, min_, max_)" % inputs[0])
if onnx_opset_version() >= 11:
Clip = Clip_11
else:
Clip = Clip_6
