"""
    pint.facets.context.objects
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~

    :copyright: 2022 by Pint Authors, see AUTHORS for more details.
    :license: BSD, see LICENSE for more details.
"""

from __future__ import annotations

import weakref
from collections import ChainMap, defaultdict
from collections.abc import Callable, Iterable
from typing import TYPE_CHECKING, Any, Generic, Protocol

from ..._typing import Magnitude
from ...facets.plain import MagnitudeT, PlainQuantity, PlainUnit, UnitDefinition
from ...util import UnitsContainer, to_units_container
from .definitions import ContextDefinition

if TYPE_CHECKING:
    from ...registry import UnitRegistry


class Transformation(Protocol):
    def __call__(
        self, ureg: UnitRegistry, value: PlainQuantity, **kwargs: Any
    ) -> PlainQuantity:
        ...


from ..._typing import UnitLike

ToBaseFunc = Callable[[UnitsContainer], UnitsContainer]
SrcDst = tuple[UnitsContainer, UnitsContainer]


class ContextQuantity(Generic[MagnitudeT], PlainQuantity[MagnitudeT]):
    pass


class ContextUnit(PlainUnit):
    pass


class Context:
    """A specialized container that defines transformation functions from one
    dimension to another. Each Dimension are specified using a UnitsContainer.
    Simple transformation are given with a function taking a single parameter.

    Conversion functions may take optional keyword arguments and the context
    can have default values for these arguments.

    Additionally, a context may host redefinitions.

    A redefinition must be performed among units that already exist in the registry. It
    cannot change the dimensionality of a unit. The symbol and aliases are automatically
    inherited from the registry.

    See ContextDefinition for the definition file syntax.

    Parameters
    ----------
    name : str or None, optional
        Name of the context (must be unique within the registry).
        Use None for anonymous Context. (Default value = None).
    aliases : iterable of str
        Other names for the context.
    defaults : None or dict
        Maps variable names to values.

    Example
    -------

    >>> from pint.util import UnitsContainer
    >>> from pint import Context, UnitRegistry
    >>> ureg = UnitRegistry()
    >>> timedim = UnitsContainer({'[time]': 1})
    >>> spacedim = UnitsContainer({'[length]': 1})
    >>> def time_to_len(ureg, time):
    ...     'Time to length converter'
    ...     return 3. * time
    >>> c = Context()
    >>> c.add_transformation(timedim, spacedim, time_to_len)
    >>> c.transform(timedim, spacedim, ureg, 2)
    6.0
    >>> def time_to_len_indexed(ureg, time, n=1):
    ...     'Time to length converter, n is the index of refraction of the material'
    ...     return 3. * time / n
    >>> c = Context(defaults={'n':3})
    >>> c.add_transformation(timedim, spacedim, time_to_len_indexed)
    >>> c.transform(timedim, spacedim, ureg, 2)
    2.0
    >>> c.redefine("pound = 0.5 kg")
    """

    def __init__(
        self,
        name: str | None = None,
        aliases: tuple[str, ...] = tuple(),
        defaults: dict[str, Any] | None = None,
    ) -> None:
        self.name: str | None = name
        self.aliases: tuple[str, ...] = aliases

        #: Maps (src, dst) -> transformation function
        self.funcs: dict[SrcDst, Transformation] = {}

        #: Maps defaults variable names to values
        self.defaults: dict[str, Any] = defaults or {}

        # Store Definition objects that are context-specific
        # TODO: narrow type this if possible.
        self.redefinitions: list[Any] = []

        # Flag set to True by the Registry the first time the context is enabled
        self.checked = False

        #: Maps (src, dst) -> self
        #: Used as a convenience dictionary to be composed by ContextChain
        self.relation_to_context: weakref.WeakValueDictionary[
            SrcDst, Context
        ] = weakref.WeakValueDictionary()

    @classmethod
    def from_context(cls, context: Context, **defaults: Any) -> Context:
        """Creates a new context that shares the funcs dictionary with the
        original context. The default values are copied from the original
        context and updated with the new defaults.

        If defaults is empty, return the same context.

        Parameters
        ----------
        context : pint.Context
            Original context.
        **defaults


        Returns
        -------
        pint.Context
        """
        if defaults:
            newdef = dict(context.defaults, **defaults)
            c = cls(context.name, context.aliases, newdef)
            c.funcs = context.funcs
            c.redefinitions = context.redefinitions
            for edge in context.funcs:
                c.relation_to_context[edge] = c
            return c
        return context

    @classmethod
    def from_lines(
        cls,
        lines: Iterable[str],
        to_base_func: ToBaseFunc | None = None,
        non_int_type: type = float,
    ) -> Context:
        context_definition = ContextDefinition.from_lines(lines, non_int_type)

        if context_definition is None:
            raise ValueError(f"Could not define Context from from {lines}")

        return cls.from_definition(context_definition, to_base_func)

    @classmethod
    def from_definition(
        cls, cd: ContextDefinition, to_base_func: ToBaseFunc | None = None
    ) -> Context:
        ctx = cls(cd.name, cd.aliases, cd.defaults)

        for definition in cd.redefinitions:
            ctx._redefine(definition)

        for relation in cd.relations:
            try:
                # TODO: check to_base_func. Is it a good API idea?
                if to_base_func:
                    src = to_base_func(relation.src)
                    dst = to_base_func(relation.dst)
                else:
                    src, dst = relation.src, relation.dst
                ctx.add_transformation(src, dst, relation.transformation)
                if relation.bidirectional:
                    ctx.add_transformation(dst, src, relation.transformation)
            except Exception as exc:
                raise ValueError(
                    f"Could not add Context {cd.name} relation {relation}"
                ) from exc

        return ctx

    def add_transformation(
        self, src: UnitLike, dst: UnitLike, func: Transformation
    ) -> None:
        """Add a transformation function to the context."""

        _key = self.__keytransform__(src, dst)
        self.funcs[_key] = func
        self.relation_to_context[_key] = self

    def remove_transformation(self, src: UnitLike, dst: UnitLike) -> None:
        """Add a transformation function to the context."""

        _key = self.__keytransform__(src, dst)
        del self.funcs[_key]
        del self.relation_to_context[_key]

    @staticmethod
    def __keytransform__(src: UnitLike, dst: UnitLike) -> SrcDst:
        return to_units_container(src), to_units_container(dst)

    def transform(
        self, src: UnitLike, dst: UnitLike, registry: Any, value: Magnitude
    ) -> Magnitude:
        """Transform a value."""

        _key = self.__keytransform__(src, dst)
        func = self.funcs[_key]
        return func(registry, value, **self.defaults)

    def redefine(self, definition: str) -> None:
        """Override the definition of a unit in the registry.

        Parameters
        ----------
        definition : str
            <unit> = <new definition>``, e.g. ``pound = 0.5 kg``
        """
        from ...delegates import ParserConfig, txt_defparser

        # TODO: kept for backwards compatibility.
        #       this is not a good idea as we have no way of known the correct non_int_type
        cfg = ParserConfig(float)
        parser = txt_defparser.DefParser(cfg, None)
        pp = parser.parse_string(definition)
        for definition in parser.iter_parsed_project(pp):
            if isinstance(definition, UnitDefinition):
                self._redefine(definition)

    def _redefine(self, definition: UnitDefinition):
        self.redefinitions.append(definition)

    def hashable(
        self,
    ) -> tuple[
        str | None,
        tuple[str, ...],
        frozenset[tuple[SrcDst, int]],
        frozenset[tuple[str, Any]],
        tuple[Any, ...],
    ]:
        """Generate a unique hashable and comparable representation of self, which can
        be used as a key in a dict. This class cannot define ``__hash__`` because it is
        mutable, and the Python interpreter does cache the output of ``__hash__``.

        Returns
        -------
        tuple
        """
        return (
            self.name,
            tuple(self.aliases),
            frozenset((k, id(v)) for k, v in self.funcs.items()),
            frozenset(self.defaults.items()),
            tuple(self.redefinitions),
        )


class ContextChain(ChainMap[SrcDst, Context]):
    """A specialized ChainMap for contexts that simplifies finding rules
    to transform from one dimension to another.
    """

    def __init__(self):
        super().__init__()
        self.contexts: list[Context] = []
        self.maps.clear()  # Remove default empty map
        self._graph: dict[SrcDst, set[UnitsContainer]] | None = None

    def insert_contexts(self, *contexts: Context):
        """Insert one or more contexts in reversed order the chained map.
        (A rule in last context will take precedence)

        To facilitate the identification of the context with the matching rule,
        the *relation_to_context* dictionary of the context is used.
        """

        self.contexts = list(reversed(contexts)) + self.contexts
        self.maps = [ctx.relation_to_context for ctx in reversed(contexts)] + self.maps
        self._graph = None

    def remove_contexts(self, n: int | None = None):
        """Remove the last n inserted contexts from the chain.

        Parameters
        ----------
        n: int
            (Default value = None)
        """

        del self.contexts[:n]
        del self.maps[:n]
        self._graph = None

    @property
    def defaults(self) -> dict[str, Any]:
        for ctx in self.values():
            return ctx.defaults
        return {}

    @property
    def graph(self):
        """The graph relating"""
        if self._graph is None:
            self._graph = defaultdict(set)
            for fr_, to_ in self:
                self._graph[fr_].add(to_)
        return self._graph

    # TODO: type registry
    def transform(
        self, src: UnitsContainer, dst: UnitsContainer, registry: Any, value: Magnitude
    ):
        """Transform the value, finding the rule in the chained context.
        (A rule in last context will take precedence)
        """
        return self[(src, dst)].transform(src, dst, registry, value)

    def hashable(self) -> tuple[Any, ...]:
        """Generate a unique hashable and comparable representation of self, which can
        be used as a key in a dict. This class cannot define ``__hash__`` because it is
        mutable, and the Python interpreter does cache the output of ``__hash__``.
        """
        return tuple(ctx.hashable() for ctx in self.contexts)