pre_processing.py

__author__ = "Lukas Leufen, Felix Kleinert"
__date__ = '2019-11-25'


import logging
from typing import Tuple, Dict, List

from src.data_handling.data_generator import DataGenerator
from src.helpers import TimeTracking
from src.join import EmptyQueryResult
from src.run_modules.run_environment import RunEnvironment

DEFAULT_ARGS_LIST = ["data_path", "network", "stations", "variables", "interpolate_dim", "target_dim", "target_var"]
DEFAULT_KWARGS_LIST = ["limit_nan_fill", "window_history_size", "window_lead_time", "statistics_per_var",
                       "station_type", "overwrite_local_data", "start", "end", "sampling", "transformation"]


class PreProcessing(RunEnvironment):

    """
    Pre-process your data by using this class. It includes time tracking and uses the experiment setup to look for data
    and stores it if not already in local disk. Further, it provides this data as a generator and checks for valid
    stations (in this context: valid=data available). Finally, it splits the data into valid training, validation and
    testing subsets.
    """

    def __init__(self):

        # create run framework
        super().__init__()

        #
        self._run()

    def _run(self):
        args = self.data_store.create_args_dict(DEFAULT_ARGS_LIST, scope="general.preprocessing")
        kwargs = self.data_store.create_args_dict(DEFAULT_KWARGS_LIST, scope="general.preprocessing")
        stations = self.data_store.get("stations", "general")
        valid_stations = self.check_valid_stations(args, kwargs, stations, load_tmp=False, save_tmp=False)
        self.data_store.set("stations", valid_stations, "general")
        self.split_train_val_test()
        self.report_pre_processing()

    def report_pre_processing(self):
        logging.debug(20 * '##')
        n_train = len(self.data_store.get('generator', 'general.train'))
        n_val = len(self.data_store.get('generator', 'general.val'))
        n_test = len(self.data_store.get('generator', 'general.test'))
        n_total = n_train + n_val + n_test
        logging.debug(f"Number of all stations: {n_total}")
        logging.debug(f"Number of training stations: {n_train}")
        logging.debug(f"Number of val stations: {n_val}")
        logging.debug(f"Number of test stations: {n_test}")
        logging.debug(f"TEST SHAPE OF GENERATOR CALL: {self.data_store.get('generator', 'general.test')[0][0].shape}"
                      f"{self.data_store.get('generator', 'general.test')[0][1].shape}")

    def split_train_val_test(self) -> None:
        """
        Splits all subsets. Currently: train, val, test and train_val (actually this is only the merge of train and val,
        but as an separate generator). IMPORTANT: Do not change to order of the execution of create_set_split. The train
        subset needs always to be executed at first, to set a proper transformation.
        """
        fraction_of_training = self.data_store.get("fraction_of_training", "general")
        stations = self.data_store.get("stations", "general")
        train_index, val_index, test_index, train_val_index = self.split_set_indices(len(stations), fraction_of_training)
        subset_names = ["train", "val", "test", "train_val"]
        if subset_names[0] != "train":  # pragma: no cover
            raise AssertionError(f"Make sure, that the train subset is always at first execution position! Given subset"
                                 f"order was: {subset_names}.")
        for (ind, scope) in zip([train_index, val_index, test_index, train_val_index], subset_names):
            self.create_set_split(ind, scope)

    @staticmethod
    def split_set_indices(total_length: int, fraction: float) -> Tuple[slice, slice, slice, slice]:
        """
        create the training, validation and test subset slice indices for given total_length. The test data consists on
        (1-fraction) of total_length (fraction*len:end). Train and validation data therefore are made from fraction of
        total_length (0:fraction*len). Train and validation data is split by the factor 0.8 for train and 0.2 for
        validation. In addition, split_set_indices returns also the combination of training and validation subset.
        :param total_length: list with all objects to split
        :param fraction: ratio between test and union of train/val data
        :return: slices for each subset in the order: train, val, test, train_val
        """
        pos_test_split = int(total_length * fraction)
        train_index = slice(0, int(pos_test_split * 0.8))
        val_index = slice(int(pos_test_split * 0.8), pos_test_split)
        test_index = slice(pos_test_split, total_length)
        train_val_index = slice(0, pos_test_split)
        return train_index, val_index, test_index, train_val_index

    def create_set_split(self, index_list: slice, set_name) -> None:
        """
        Create the subset for given split index and stores the DataGenerator with given set name in data store as
        `generator`. Checks for all valid stations using the default (kw)args for given scope and creates the
        DataGenerator for all valid stations. Also sets all transformation information, if subset is training set. Make
        sure, that the train set is executed first, and all other subsets afterwards.
        :param index_list: list of all stations to use for the set. If attribute use_all_stations_on_all_data_sets=True,
            this list is ignored.
        :param set_name: name to load/save all information from/to data store without the leading general prefix.
        """
        scope = f"general.{set_name}"
        args = self.data_store.create_args_dict(DEFAULT_ARGS_LIST, scope)
        kwargs = self.data_store.create_args_dict(DEFAULT_KWARGS_LIST, scope)
        stations = args["stations"]
        if self.data_store.get("use_all_stations_on_all_data_sets", scope):
            set_stations = stations
        else:
            set_stations = stations[index_list]
        logging.debug(f"{set_name.capitalize()} stations (len={len(set_stations)}): {set_stations}")
        set_stations = self.check_valid_stations(args, kwargs, set_stations, load_tmp=False)
        self.data_store.set("stations", set_stations, scope)
        set_args = self.data_store.create_args_dict(DEFAULT_ARGS_LIST, scope)
        data_set = DataGenerator(**set_args, **kwargs)
        self.data_store.set("generator", data_set, scope)
        if set_name == "train":
            self.data_store.set("transformation", data_set.transformation, "general")

    @staticmethod
    def check_valid_stations(args: Dict, kwargs: Dict, all_stations: List[str], load_tmp=True, save_tmp=True):
        """
        Check if all given stations in `all_stations` are valid. Valid means, that there is data available for the given
        time range (is included in `kwargs`). The shape and the loading time are logged in debug mode.
        :param args: Dictionary with required parameters for DataGenerator class (`data_path`, `network`, `stations`,
            `variables`, `interpolate_dim`, `target_dim`, `target_var`).
        :param kwargs: positional parameters for the DataGenerator class (e.g. `start`, `interpolate_method`,
            `window_lead_time`).
        :param all_stations: All stations to check.
        :return: Corrected list containing only valid station IDs.
        """
        t_outer = TimeTracking()
        t_inner = TimeTracking(start=False)
        logging.info("check valid stations started")
        valid_stations = []

        # all required arguments of the DataGenerator can be found in args, positional arguments in args and kwargs
        data_gen = DataGenerator(**args, **kwargs)
        for station in all_stations:
            t_inner.run()
            try:
                # (history, label) = data_gen[station]
                data = data_gen.get_data_generator(key=station, load_local_tmp_storage=load_tmp,
                                                   save_local_tmp_storage=save_tmp)
                valid_stations.append(station)
                logging.debug(f'{station}: history_shape = {data.history.transpose("datetime", "window", "Stations", "variables").shape}')
                logging.debug(f"{station}: loading time = {t_inner}")
            except (AttributeError, EmptyQueryResult):
                continue
        logging.info(f"run for {t_outer} to check {len(all_stations)} station(s). Found {len(valid_stations)}/"
                     f"{len(all_stations)} valid stations.")
        return valid_stations