sasctl.utils.pymas package#

Submodules#

sasctl.utils.pymas.core module#

Contains utilities for wrapping Python models in DS2 for publishing.

class sasctl.utils.pymas.core.PyMAS(target_function, variables, python_source, return_code=None, return_msg=None, func_prefix=None, **kwargs)[source]#

Bases: object

Parameters:

  • target_function (str) – The Python function to be executed.

  • variables (list of DS2Variable) – The input/ouput variables be declared in the module.

  • python_source (str) – Additional Python code to be executed during setup.

  • return_code (bool) – Deprecated. Whether the DS2-generated return code should be included.

  • return_msg (bool) – Deprecated. Whether the DS2-generated return message should be included.

  • **kwargs – Passed to build_wrapper_function().

score_code(input_table=None, output_table=None, columns=None, dest='MAS')[source]#

Generate DS2 score code

Parameters:

  • input_table (str) – The name of the table to execute the function against

  • output_table (str) – The name of the table where execution results will be written

  • columns (list of str) – Names of the columns from table that will be passed to func

  • dest (str) – Choose from {'MAS', 'EP', 'CAS', 'Python'}. Specifies the publishing destination for the score code to ensure that compatible code is generated.

Returns:

str – Score code

Changed in version 1.4: Added dest=’Python’ option

sasctl.utils.pymas.core.build_wrapper_function(func, variables, array_input, name='wrapper', setup=None, return_msg=None)[source]#

Wraps a function to ensure compatibility when called by PyMAS.

PyMAS has strict expectations regarding the format of any function called directly by PyMAS. Isolating the desired function inside a wrapping function provides a simple way to ensure that functions called by PyMAS are compliant.

Parameters:

  • func (Callable or str) – Function name or an instance of Function which will be wrapped

  • variables (list of DS2Variable)

  • array_input (bool) – Whether variables should be combined into a single array before passing to func

  • name (str, optional) – Name of the generated wrapper function. Defaults to ‘wrapper’.

  • setup (Iterable) – Python source code lines to be executed during package setup

  • return_msg (bool, optional) – Deprecated. Whether error messages will be captured and returned as an additional output variable. Defaults to True.

Returns:

str – The Python definition for the wrapper function.

Notes

The format for the `# Output: ` is very strict. It must be exactly “# Output: <var>, <var>”. Any changes to spelling, capitalization, punctuation, or spacing will result in an error when the DS2 code is executed.

Changed in version 1.5: Added name parameter.

sasctl.utils.pymas.core.from_inline(func, input_types=None, array_input=False, return_code=None, return_message=None)[source]#

Creates a PyMAS wrapper to execute the inline python function.

Parameters:

  • func (Callable) – A Python function object to be used

  • input_types (list of type, optional) – The expected type for each input value of func. Can be ommitted if func includes type hints.

  • array_input (bool) – Whether the function inputs should be treated as an array instead of individual parameters

  • return_code (bool) – Deprecated. Whether the DS2-generated return code should be included

  • return_message (bool) – Deprecated. Whether the DS2-generated return message should be included

Returns:

PyMAS – Generated DS2 code which can be executed in a SAS scoring environment

Changed in version 1.5: Return code and message are disabled by default.

sasctl.utils.pymas.core.from_pickle(file, func_name=None, input_types=None, array_input=False, return_code=None, return_message=None)[source]#

Create a deployable DS2 package from a Python pickle file.

Parameters:

  • file (str or bytes or io.BytesIO) – Pickled object to use. String is assumed to be a path to a picked file, file_like is assumed to be an open file handle to a pickle object, and bytes is assumed to be the raw pickled bytes.

  • func_name (str) – Name of the target function to call

  • input_types (pandas.DataFrame, type, list of type, or dict of str, optional) – The expected type for each input value of the target function. Can be omitted if target function includes type hints. If a DataFrame is provided, the columns will be inspected to determine type information. If a single type is provided, all columns will be assumed to be that type, otherwise a list of column types or a dictionary of column_name: type may be provided.

  • array_input (bool) – Whether the function inputs should be treated as an array instead of individual parameters

  • return_code (bool) – Deprecated. Whether the DS2-generated return code should be included

  • return_message (bool) – Deprecated. Whether the DS2-generated return message should be included

Returns:

PyMAS – Generated DS2 code which can be executed in a SAS scoring environment

Changed in version 1.5: Return code and message are disabled by default.

sasctl.utils.pymas.core.from_python_file(file, func_name=None, input_types=None, array_input=False, return_code=None, return_message=None)[source]#

Creates a PyMAS wrapper to execute a function defined in an external .py file.

Parameters:

  • file (str) – The path to a python source code file

  • func_name (str) – Name of the target function to call

  • input_types (list of type, optional) – The expected type for each input value of the target function. Can be ommitted if target function includes type hints.

  • array_input (bool) – Whether the function inputs should be treated as an array instead of individual parameters

  • return_code (bool) – Deprecated Whether the DS2-generated return code should be included

  • return_message (bool) – Deprecated Whether the DS2-generated return message should be included

Returns:

PyMAS – Generated DS2 code which can be executed in a SAS scoring environment

Changed in version 1.5: Return code and message are disabled by default.

sasctl.utils.pymas.core.wrap_predict_method(func, variables, **kwargs)[source]#

Create a PyMAS wrapper designed for Sci-kit’s .predict methods.

Parameters:

  • func (Callable or str) – Function name or an instance of Function which will be wrapped. Assumed to behave as .predict() methods.

  • variables (list of DS2Variable) – Input and output variables for the function

  • **kwargs – Will be passed to build_wrapper_function().

Returns:

str – the Python definition for the wrapper function.

See also

build_wrapper_function

Added in version 1.5.

sasctl.utils.pymas.core.wrap_predict_proba_method(func, variables, **kwargs)[source]#

Create a PyMAS wrapper designed for Sci-kit’s .predict_proba methods.

Parameters:

  • func (Callable or str) – Function name or an instance of Function which will be wrapped. Assumed to behave as .predict_proba() methods.

  • variables (list of DS2Variable) – Input and output variables for the function

  • **kwargs – Will be passed to build_wrapper_function.

Returns:

str – the Python definition for the wrapper function.

See also

build_wrapper_function

Added in version 1.5.

sasctl.utils.pymas.ds2 module#

The ds2 module contains classes representing DS2 code constructs.

class sasctl.utils.pymas.ds2.DS2BaseMethod(name, variables, body=None)[source]#

Bases: object

code()[source]#
property name#
class sasctl.utils.pymas.ds2.DS2BasePackage(code=None)[source]#

Bases: object

Defines a DS2 package.

Parameters:

code (str) – Any code to include in the body of the package.

Variables:

methods (list) – A collection of DS2BaseMethod instances that will be included in the package definition.

Added in version 1.5.

code()[source]#

Get the DS2 code for the package.

Returns:

str

property id#

Unique identifier generated for the package.

property name#

Unique name generated for the package.

sasctl.utils.pymas.ds2.DS2Method(variables, code, target='wrapper')[source]#

Deprecated since version 1.5.

sasctl.utils.pymas.ds2.DS2Package(variables, code=None, return_code=True, return_message=True, target=None)[source]#

Deprecated since version 1.5: Use DS2PyMASPackage instead.

class sasctl.utils.pymas.ds2.DS2PyMASMethod(name, variables, python_code, return_code=None, return_message=None, target='wrapper', method_name='score')[source]#

Bases: DS2BaseMethod

A DS2 method that builds a PyMAS object.

Parameters:

  • name (str) – Name of Python model to define

  • variables

  • python_code

  • return_code

  • return_message

  • target

  • method_name

class sasctl.utils.pymas.ds2.DS2PyMASPackage(code=None)[source]#

Bases: DS2BasePackage

A DS2 package that uses PyMAS to invoke Python code.

Parameters:

code

Added in version 1.5.

add_method(name, target, variables)[source]#

Add a DS2 method that calls a Python function defined by the package.

Parameters:

  • name (str) – Name of the DS2 method to create.

  • target (str) – Name of the Python method to call

  • variables (list of DS2Variable) – List of input and output variables for the method.

Returns:

None

sasctl.utils.pymas.ds2.DS2ScoreMethod(variables, return_code=True, return_message=True, target='wrapper')[source]#

Deprecated since version 1.5.

class sasctl.utils.pymas.ds2.DS2Thread(variables, table, column_names=None, return_code=True, return_message=True, package=None, method=None)[source]#

Bases: object

property id#
property name#
class sasctl.utils.pymas.ds2.DS2Variable(*args, **kwargs)[source]#

Bases: Ds2Variable

DS2_TYPE_TO_VIYA = {'char': 'string', 'double': 'decimal', 'varchar': 'string'}#
PY_TYPE_TO_DS2 = {'double': 'double', 'double32': 'double', 'double64': 'double', 'float': 'double', 'float32': 'double', 'float64': 'double', 'int': 'integer', 'int32': 'integer', 'int64': 'integer', 'integer': 'integer', 'integer32': 'integer', 'integer64': 'integer', 'str': 'char', 'string': 'char', 'uint16': 'integer', 'uint32': 'integer', 'uint64': 'integer', 'uint8': 'integer', 'varchar': 'char'}#
as_declaration()[source]#

DS2 variable declaration statement.

as_model_metadata()[source]#
as_parameter()[source]#

DS2 parameter syntax for method signatures.

property is_array#
pymas_statement(python_var_name=None)[source]#

Returns appropriate PyMAS get/set statements to move values between a Python variable and the DS2 variable.

Parameters:

python_var_name (str) – Python variable name.

Raises:

ValueError –

property size#

sasctl.utils.pymas.python module#

sasctl.utils.pymas.python.ds2_variables(input, output_vars=False, names=None)[source]#

Generate a collection of DS2Variable instances corresponding to the input

Parameters:

  • input (Callable or OrderedDict[str, tuple] or OrderedDict[str, type] or pandas.DataFrame) – a function or mapping parameter names to (type, is_output)

  • output_vars (bool) – Whether or not to treat all variables from input as output variables

  • names (list of str) – List of variable names to used. If a single string is specified it will be used as a prefix and variable names in the format “prefixNN” will be generated.

Returns:

list of DS2Variable

Examples

>>> ds2_variables(OrderedDict(a=int, c=float))
[DS2Variable(name='a', type='integer', out=False), DS2Variable(name='c', type='double', out=False)]

>>> ds2_variables({'x': (float, True)})
[DS2Variable(name='x', type='double', out=True)]

Changed in version 1.5: names parameter added.

sasctl.utils.pymas.python.parse_type_hints(func, skip_var='self')[source]#

Attempt to discern types for the input and output variable(s).

DS2 is a strongly-typed language but Python is not. Need to determine the types for input/output variables so they can be correctly mapped to DS2 types.

Parameters:

  • func (Callable) – the object to inspect for parameters

  • skip_var (str) – name of the variable assumed to be the instance reference, if any. Will be ignored during parameter parsing.

Returns:

params (OrderedDict) – dictionary of str : (str, bool) mapping param_name : (param_type, is_return_val)

Raises:

ValueError – If the type of one or more variables could not be determined.

sasctl.utils.pymas.python.parse_type_hints_from_annotations(func, skip_var='self')[source]#

Parse type hints from the function signature.

sasctl.utils.pymas.python.parse_type_hints_from_source(func, skip_var='self')[source]#

Parse type hints stored in comments according to PEP 484.

Module contents#