Logging functions

LoggingFunction base class

To develop custom loggers, we need to write a class that inherits from the base class LoggingFunction. Then, those custom loggers can be mapped to events of the user choice in the parameter mapping of the PyPads class.

class pypads.functions.loggers.base_logger.LoggingFunction(*args, static_parameters=None, **kwargs)

This class should be used to define new custom loggers. The user has to define __pre__ and/or __post__ methods depending on the specific use case.

Parameters:static_parameters – dict, optional, static parameters (if needed) to be used when logging.

Note

It is not recommended to change the __call_wrapped__ method, only if really needed.

__call_wrapped__(ctx, *args, _pypads_env: pypads.functions.analysis.call_tracker.LoggingEnv, _args, _kwargs, **_pypads_hook_params)

The real call of the wrapped function. Be carefull when you change this. Exceptions here will not be catched automatically and might break your workflow. The returned value will be passed on to __post__ function.

Returns:_pypads_result

__post__(ctx, *args, _pypads_env, _pypads_pre_return, _pypads_result, _args, _kwargs, **kwargs)

The function to be called after executing the log anchor.

Parameters:

• _pypads_pre_return – the value returned by __pre__.
• _pypads_result – the value returned by __call_wrapped__.

Returns:

the wrapped function return value

__pre__(ctx, *args, _pypads_env, _args, _kwargs, **kwargs)

The function to be called before executing the log anchor. the value returned will be passed on to the __post__ function as _pypads_pre_return.

Returns:_pypads_pre_return

_check_dependencies()

Raise error if dependencies are missing.

static _needed_packages()

Overwrite this to provide your package names. :return: List of needed packages by the logger.

Pre and Post run loggers

Another type of logging functions supported by Pypads is the pre/post run loggers which are executed before and after the run execution respectively.

• Pre Run loggers

class pypads.functions.pre_run.pre_run.PreRunFunction(*args, fn=None, **kwargs)

This class should be used to define new pre run functions

_call(pads, *args, **kwargs)

Function where to add you custom code to execute before starting the run.

Parameters:pads – the current instance of PyPads.

_check_dependencies()

Raise error if dependencies are missing.

static _needed_packages()

Overwrite this to provide your package names. :return: List of needed packages by the logger.

• Post Run loggers

class pypads.functions.post_run.post_run.PostRunFunction(*args, fn=None, message=None, **kwargs)

This class should be used to define new post run functions

_call(pads, *args, **kwargs)

Function where to add you custom code to execute after ending the run.

Parameters:pads – the current instance of PyPads.

_check_dependencies()

Raise error if dependencies are missing.

static _needed_packages()

Overwrite this to provide your package names. :return: List of needed packages by the logger.

Mlflow autolog (experimental)

Pypads also support mlflow autologging functionalities. More on that can be found at MLflow.

class pypads.functions.loggers.mlflow.mlflow_autolog.MlflowAutologger(*args, order=-1, **kwargs)

MlflowAutologger is the intergration of the mlflow autologging functionalities into PyPads tracking system.

__call_wrapped__(ctx, *args, _args, _kwargs, _pypads_autologgers=None, _pypads_env=<class 'pypads.functions.analysis.call_tracker.LoggingEnv'>, **kwargs)

Note

Experimental: This method may change or be removed in a future release without warning.

Function used to enable autologgers of mlflow.