data0.run_functions
===================

.. py:module:: data0.run_functions


Functions
---------

.. autoapisummary::

   data0.run_functions.process_cmd_args
   data0.run_functions.make_output_metadata_dict
   data0.run_functions.prepare_input


Module Contents
---------------

.. py:function:: process_cmd_args(cmd_args, verbose=True, default_savedir='HPC_runs/test_unet0', default_config='input_config', default_version=1)

   Process command line arguments given to `run_model.py`.

   :param cmd_args: The list of command line arguments given to `run_model.py`.
                    Assumes the arguments are:
                    - arg 0: script name (i.e., `run_model.py`)
                    - arg 1: savedir, the directory in which to save model outputs
                    - arg 2: version, the version of the packages to import (0 or 1)
   :type cmd_args: `list`
   :param verbose: Whether to print the processed command line arguments.
                   Default is True.
   :type verbose: `bool`, optional
   :param default_savedir: The default save directory to use if none is found in `cmd_args`.
                           Default is 'HPC_runs/test_unet'.
   :type default_savedir: `str`, optional
   :param default_config: The default config file to use if none is found in `cmd_args`.
                          Default is 'input_config'.
   :type default_config: `str`, optional
   :param default_version: The default version to use if none is found in `cmd_args`.
                           Default is 1.
   :type default_version: `int`, optional

   :returns: * **savedir** (`str`) -- The path to the directory in which to save model outputs.
             * **config_dict** (`dict`) -- The model configuration dictionary.
             * **config_path** (`str`) -- The path to the configuration file used.
             * **version** (`int`) -- The version of the packages to use for running the model (0 or 1).


.. py:function:: make_output_metadata_dict(savedir, config_path, config_dict, version, model_fmt)

   Create the dictionary of metadata for a run to be output to a dictionary.

   :param savedir: The path to the directory in which the data for this run is saved.
   :type savedir: `str`
   :param config_path: The path to the configuration JSON file.
   :type config_path: `str`
   :param config_dict: The dictionary of the configuration file.
   :type config_dict: `dict`
   :param version: The version of the code used in this run (0 or 1).
   :type version: `int`
   :param model_fmt: The format in which to output the trained model for this run.
                     Either 'h5', 'keras', or 'both'.
   :type model_fmt: `str`

   :returns: **output_metadata** -- The output metadata dictionary.
   :rtype: `dict`


.. py:function:: prepare_input(uarr, input_config, output_metadata, split_year=2019, stage=1)

   Prepare the input data for the model.

   Get the training data from the input NetCDF dataset as numpy arrays and concatenate them along the time dimension.

   :param uarr: The dataset of the input NetCDF file.
   :type uarr: `unox.uarray`
   :param input_config: Path to the input configuration JSON file or a dictionary containing the configuration.
   :type input_config: `str` or `dict`
   :param output_metadata: The dictionary of metadata describing the output of a model run.
   :type output_metadata: `dict`
   :param split_year: The year at which to split the training and testing data.
                      Defaults to 2019.
   :type split_year: `int`, optional
   :param stage: The stage of the data to plot (1 or 2).
   :type stage: `int`

   :returns: * **xtrain** (`np.ndarray`) -- Concatenated training input features.
             * **ytrain** (`np.ndarray`) -- Concatenated training target variables.
             * **output_metadata** (`dict`) -- The dictionary of metadata describing the output of a model run with values added for `train_years` and `unet_build_shape`.