Clean slate for Forecast

Author: peanutfun

climada/util/forecast.py

@@ -19,117 +19,6 @@
 Define Forecast base class.
 """
 
-from abc import ABC, abstractmethod
-from typing import Literal, Self
 
-import numpy as np
-from scipy.sparse import block_array, csr_matrix
-
-
-def reduce_mat(mat: csr_matrix, func):
-    """Reduce a matrix and return the CSR representation"""
-    return func(mat).tocsr()
-
-
-def concat_matrices_per_event(*matrices: csr_matrix):
-    """Concatenate matrices by event"""
-    return block_array([[mat] for mat in matrices], format="csr")
-
-
-def matrix_event_padding(mat: csr_matrix, num_events):
-    """Pad zero events"""
-    pad_events = mat.shape[0] - num_events
-    if pad_events < 1:
-        return mat
-    return block_array(
-        [[mat], csr_matrix((pad_events, mat.shape[1], mat.dtype))], format="csr"
-    )
-
-
-class Forecast(ABC):
-    lead_time: np.ndarray[np.timedelta64]
-    member: np.ndarray[int]
-    forecast_date: np.datetime64 | None
-    num_members: int
-    num_lead_times: int
-
-    def __init__(self, lead_time, member, forecast_date: np.datetime64 | None = None):
-        """Store members"""
-        pass
-
-    # --- Selection --- #
-
-    @abstractmethod
-    def _select_by_index(self, index: tuple[np.ndarray, ...]) -> Self:
-        """Return a new object with the index used for selecting events"""
-        raise NotImplementedError
-
-    def _select_member(self, member: int | None) -> np.ndarray:
-        """Return boolean array where self.member == member"""
-        ...
-
-    def _select_lead_time(self, lead_time: np.timedelta64 | None) -> np.ndarray:
-        """Return boolean array where self.lead_time == lead_time"""
-        ...
-
-    def select(self, *, member: int | None, lead_time: np.timedelta64 | None) -> Self:
-        index = np.nonzero(
-            self._select_member(member) & self._select_lead_time(lead_time)
-        )
-        return self._select_by_index(index)
-
-    # --- Generic reduction --- #
-
-    @classmethod
-    @abstractmethod
-    def concat(cls, *obj: Self) -> Self:
-        """Concatenate multiple object instances"""
-        raise NotImplementedError
-
-    @abstractmethod
-    def _reduce(self, func) -> Self:
-        """Apply the reduction function in the derived class and return the result
-
-        Note: The derived class will likely need to pad matrices!
-        """
-        raise NotImplementedError
-
-    def reduce(self, func, dim: Literal["member", "lead_time"] | None = None) -> Self:
-        """Reduce along a given dimension with func"""
-        if dim is None:
-            # TODO: Check if we selected a specific member or lead time.
-            #       Pad events accordingly!
-            return self._reduce(func=func)  # Derived class specialization
-
-        return self.concat(
-            *(
-                self.select(**{dim: val}).reduce(func=func, dim=None)
-                for val in np.unique(getattr(self, dim))
-            )
-        )
-
-    # --- Specializations --- #
-
-    @abstractmethod
-    def _max(self) -> Self:
-        """Apply the maximum function in the derived class and return the result"""
-        raise NotImplementedError
-
-    def _reduce_attr(
-        self, attr: str, dim: Literal["member", "lead_time"] | None = None
-    ) -> Self:
-        """Reduce along a given dimension with attribute attr"""
-        if dim is None:
-            # TODO: Check if we selected a specific member or lead time.
-            #       Pad events accordingly!
-            return getattr(self, "_" + attr)()  # Derived class specialization
-
-        return self.concat(
-            *(
-                getattr(self.select(**{dim: val}), attr)(dim=None)
-                for val in np.unique(getattr(self, dim))
-            )
-        )
-
-    def max(self, dim):
-        return self._reduce_attr(attr="max", dim=dim)
+class Forecast:
+    pass