module `onnx_tools.onnx2py_helper`#

Short summary#

module mlprodict.onnx_tools.onnx2py_helper

Functions which converts ONNX object into readable python objects.

Functions#

function	truncated documentation
`_elem_type_as_str`
`_get_onnx_function`	Returns the list of functions defined in ONNX package.
`_numpy_array`	Single function to create an array.
`_sparse_array`	Single function to create an sparse array (coo_matrix).
`_to_array`
`_type_to_string`	Converts a type into a readable string.
`_var_as_dict`	Converts a protobuf object into something readable. The current implementation relies on json. That’s not …
`copy_value_info`	Makes a copy of onnx.ValueInfoProto.
`from_array`	Converts an array into an ONNX tensor.
`from_bytes`	Retrieves an array from bytes then protobuf.
`from_pb`	Extracts tensor description from a protobuf.
`get_dtype_shape`	Returns the shape of a tensor.
`get_onnx_schema`	Returns the operator schema for a specific operator.
`get_tensor_elem_type`	Returns the element type if that makes sense for this object.
`get_tensor_shape`	Returns the shape if that makes sense for this object.
`guess_dtype`	Converts a proto type into a numpy type.
`guess_numpy_type_from_dtype`	Converts a string (such as ‘dtype(float32)’) into a numpy dtype.
`guess_numpy_type_from_string`	Converts a string (such as ‘float’) into a numpy dtype.
`guess_proto_dtype`	Guesses the ONNX dtype given a numpy dtype.
`guess_proto_dtype_name`	Returns a string equivalent to onnx_dtype.
`make_value_info`	Converts a variable defined by its name, type and shape into onnx.ValueInfoProto.
`numpy_max`	Returns the maximum of an array. Deals with text as well.
`numpy_min`	Returns the minimum of an array. Deals with text as well.
`numpy_type_prototype`	Converts a numpy dtyp into a TensorProto dtype.
`onnx_model_opsets`	Extracts opsets in a dictionary.
`to_bytes`	Converts an array into protobuf and then into bytes.
`to_skl2onnx_type`	Converts name, elem_type, shape into a sklearn-onnx type.

Documentation#

Functions which converts ONNX object into readable python objects.

source on GitHub

mlprodict.onnx_tools.onnx2py_helper._elem_type_as_str(elem_type)#

mlprodict.onnx_tools.onnx2py_helper._get_onnx_function()#

Returns the list of functions defined in ONNX package.

source on GitHub

mlprodict.onnx_tools.onnx2py_helper._numpy_array(data, dtype=None, copy=True)#

Single function to create an array.

Parameters:

data – data
dtype – dtype
copy – copy

Returns:

numpy array

source on GitHub

mlprodict.onnx_tools.onnx2py_helper._sparse_array(shape, data, indices, dtype=None, copy=True)#

Single function to create an sparse array (coo_matrix).

Parameters:

shape – shape
data – data
indices – indices
dtype – dtype
copy – copy

Returns:

coo_matrix

source on GitHub

mlprodict.onnx_tools.onnx2py_helper._to_array(var)#

mlprodict.onnx_tools.onnx2py_helper._type_to_string(dtype)#

Converts a type into a readable string.

source on GitHub

mlprodict.onnx_tools.onnx2py_helper._var_as_dict(var)#

Converts a protobuf object into something readable. The current implementation relies on json. That’s not the most efficient way.

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.copy_value_info(info, name=None)#

Makes a copy of onnx.ValueInfoProto.

Parameters:: name – if defined, changed the name
Returns:: instance of onnx.ValueInfoProto

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.from_array(value, name=None)#

Converts an array into an ONNX tensor.

Parameters:: value – numpy array
Returns:: ONNX tensor

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.from_bytes(b)#

Retrieves an array from bytes then protobuf.

Parameters:: b – bytes
Returns:: array

Converts bytes into an array (serialization)

Useful to deserialize.

<<<

import numpy
from mlprodict.onnx_tools.onnx2py_helper import to_bytes, from_bytes

data = numpy.array([[0, 1], [2, 3], [4, 5]], dtype=numpy.float32)
pb = to_bytes(data)
data2 = from_bytes(pb)
print(data2)

>>>

    [[0. 1.]
     [2. 3.]
     [4. 5.]]

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.from_pb(obj)#

Extracts tensor description from a protobuf.

Parameters:: obj – initializer, tensor
Returns:: (name, type, shape)

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.get_dtype_shape(obj)#

Returns the shape of a tensor.

Parameters:: obj – onnx object
Returns:: (dtype, shape) or (None, None) if not applicable

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.get_onnx_schema(opname, domain='', opset=None, load_function=False)#

Returns the operator schema for a specific operator.

Parameters:

domain – operator domain
opname – operator name
opset – opset or version, None for the latest
load_function – loads the function, if True, the function looks into the list of function if one of them has the same name, opset must be None in that case

Returns:

:epkg:`OpSchema`

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.get_tensor_elem_type(obj)#

Returns the element type if that makes sense for this object.

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.get_tensor_shape(obj)#

Returns the shape if that makes sense for this object.

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.guess_dtype(proto_type)#

Converts a proto type into a numpy type.

Parameters:: proto_type – example onnx.TensorProto.FLOAT
Returns:: numpy dtype

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.guess_numpy_type_from_dtype(dt)#

Converts a string (such as ‘dtype(float32)’) into a numpy dtype.

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.guess_numpy_type_from_string(name)#

Converts a string (such as ‘float’) into a numpy dtype.

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.guess_proto_dtype(dtype)#

Guesses the ONNX dtype given a numpy dtype.

Parameters:: dtype – numpy dtype
Returns:: proto type

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.guess_proto_dtype_name(onnx_dtype)#

Returns a string equivalent to onnx_dtype.

Parameters:: dtype – onnx dtype
Returns:: proto type

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.make_value_info(name, dtype, shape)#

Converts a variable defined by its name, type and shape into onnx.ValueInfoProto.

Parameters:

name – name
dtype – numpy element type
shape – shape

Returns:

instance of onnx.ValueInfoProto

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.numpy_max(x)#

Returns the maximum of an array. Deals with text as well.

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.numpy_min(x)#

Returns the minimum of an array. Deals with text as well.

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.numpy_type_prototype(dtype)#

Converts a numpy dtyp into a TensorProto dtype.

Parameters:: dtype – dtype
Returns:: proto dtype

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.onnx_model_opsets(onnx_model)#

Extracts opsets in a dictionary.

Parameters:: onnx_model – ONNX graph
Returns:: dictionary {domain: version}

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.to_bytes(val)#

Converts an array into protobuf and then into bytes.

Parameters:

val – array

Returns:

bytes

Converts an array into bytes (serialization)

Useful to serialize.

<<<

import numpy
from mlprodict.onnx_tools.onnx2py_helper import to_bytes

data = numpy.array([[0, 1], [2, 3], [4, 5]], dtype=numpy.float32)
pb = to_bytes(data)
print(len(pb), data.size * data.itemsize, pb[:10])

>>>

    32 24 b'\x08\x03\x08\x02\x10\x01J\x18\x00\x00'

source on GitHub

mlprodict.onnx_tools.onnx2py_helper.to_skl2onnx_type(name, elem_type, shape)#

Converts name, elem_type, shape into a sklearn-onnx type.

Parameters:

name – string
elem_type – tensor of elements of this type
shape – expected shape

Returns:

data type

source on GitHub

module onnx_tools.onnx2py_helper#

Short summary#

Functions#

Documentation#

module `onnx_tools.onnx2py_helper`#