from abc import ABCMeta, abstractmethod
from typing import ClassVar, Literal
from typing_extensions import Self

from numpy import ndarray

from .._typing import ArrayLike, Float, Int, MatrixLike
from ..base import BaseEstimator, ClassNamePrefixFeaturesOutMixin, MultiOutputMixin, RegressorMixin, TransformerMixin

__all__ = ["PLSCanonical", "PLSRegression", "PLSSVD"]

class _PLS(
    ClassNamePrefixFeaturesOutMixin,
    TransformerMixin,
    RegressorMixin,
    MultiOutputMixin,
    BaseEstimator,
    metaclass=ABCMeta,
):
    _parameter_constraints: ClassVar[dict] = ...

    @abstractmethod
    def __init__(
        self,
        n_components: int = 2,
        *,
        scale: bool = True,
        deflation_mode: str = "regression",
        mode: str = "A",
        algorithm: str = "nipals",
        max_iter: int = 500,
        tol: float = 1e-06,
        copy: bool = True,
    ) -> None: ...
    def fit(self, X: MatrixLike, Y: MatrixLike | ArrayLike) -> Self: ...
    def transform(
        self, X: MatrixLike, Y: None | MatrixLike = None, copy: bool = True
    ) -> tuple[ndarray, ...] | ndarray | tuple[ndarray, ndarray]: ...
    def inverse_transform(self, X: MatrixLike, Y: None | MatrixLike = None) -> tuple[ndarray, ndarray]: ...
    def predict(self, X: MatrixLike, copy: bool = True) -> ndarray: ...
    def fit_transform(self, X: MatrixLike, y: None | MatrixLike = None) -> ndarray: ...
    def coef_(self) -> ndarray: ...

class PLSRegression(_PLS):
    feature_names_in_: ndarray = ...
    n_features_in_: int = ...
    n_iter_: list = ...
    intercept_: ndarray = ...
    coef_: ndarray = ...
    y_rotations_: ndarray = ...
    x_rotations_: ndarray = ...
    y_scores_: ndarray = ...
    x_scores_: ndarray = ...
    y_loadings_: ndarray = ...
    x_loadings_: ndarray = ...
    y_weights_: ndarray = ...
    x_weights_: ndarray = ...

    _parameter_constraints: ClassVar[dict] = ...

    # This implementation provides the same results that 3 PLS packages
    # provided in the R language (R-project):
    #     - "mixOmics" with function pls(X, Y, mode = "regression")
    #     - "plspm " with function plsreg2(X, Y)
    #     - "pls" with function oscorespls.fit(X, Y)

    def __init__(
        self,
        n_components: Int = 2,
        *,
        scale: bool = True,
        max_iter: Int = 500,
        tol: Float = 1e-06,
        copy: bool = True,
    ) -> None: ...
    def fit(self, X: MatrixLike, Y: MatrixLike | ArrayLike) -> Self: ...

class PLSCanonical(_PLS):
    feature_names_in_: ndarray = ...
    n_features_in_: int = ...
    n_iter_: list = ...
    intercept_: ndarray = ...
    coef_: ndarray = ...
    y_rotations_: ndarray = ...
    x_rotations_: ndarray = ...
    y_loadings_: ndarray = ...
    x_loadings_: ndarray = ...
    y_weights_: ndarray = ...
    x_weights_: ndarray = ...

    _parameter_constraints: ClassVar[dict] = ...

    # This implementation provides the same results that the "plspm" package
    # provided in the R language (R-project), using the function plsca(X, Y).
    # Results are equal or collinear with the function
    # ``pls(..., mode = "canonical")`` of the "mixOmics" package. The
    # difference relies in the fact that mixOmics implementation does not
    # exactly implement the Wold algorithm since it does not normalize
    # y_weights to one.

    def __init__(
        self,
        n_components: Int = 2,
        *,
        scale: bool = True,
        algorithm: Literal["nipals", "svd"] = "nipals",
        max_iter: Int = 500,
        tol: Float = 1e-06,
        copy: bool = True,
    ) -> None: ...

class CCA(_PLS):
    feature_names_in_: ndarray = ...
    n_features_in_: int = ...
    n_iter_: list = ...
    intercept_: ndarray = ...
    coef_: ndarray = ...
    y_rotations_: ndarray = ...
    x_rotations_: ndarray = ...
    y_loadings_: ndarray = ...
    x_loadings_: ndarray = ...
    y_weights_: ndarray = ...
    x_weights_: ndarray = ...

    _parameter_constraints: ClassVar[dict] = ...

    def __init__(
        self,
        n_components: Int = 2,
        *,
        scale: bool = True,
        max_iter: Int = 500,
        tol: Float = 1e-06,
        copy: bool = True,
    ) -> None: ...

class PLSSVD(ClassNamePrefixFeaturesOutMixin, TransformerMixin, BaseEstimator):
    feature_names_in_: ndarray = ...
    n_features_in_: int = ...
    y_weights_: ndarray = ...
    x_weights_: ndarray = ...

    _parameter_constraints: ClassVar[dict] = ...

    def __init__(self, n_components: Int = 2, *, scale: bool = True, copy: bool = True) -> None: ...
    def fit(self, X: MatrixLike, Y: MatrixLike | ArrayLike) -> Self: ...
    def transform(self, X: MatrixLike, Y: None | MatrixLike | ArrayLike = None) -> ndarray | tuple[ndarray, ndarray]: ...
    def fit_transform(self, X: MatrixLike, y: None | MatrixLike | ArrayLike = None) -> ndarray | tuple[ndarray, ndarray]: ...