# BSD 3-Clause License; see https://github.com/scikit-hep/uproot4/blob/main/LICENSE

"""
This module defines the behaviors of ``TAxis``, an axis of a histogram or profile plot.
"""

from __future__ import absolute_import

try:
    from collections.abc import Sequence
except ImportError:
    from collections import Sequence

import numpy


class AxisTraits(object):
    """
    Describes read-only properties of a histogram axis.

    For example, ``axis.traits.discrete`` is True if the histogram has
    labels; False otherwise.
    """

    def __init__(self, axis):
        self._axis = axis

    def __repr__(self):
        return "AxisTraits({0})".format(repr(self._axis))

    @property
    def circular(self):
        """
        True if the axis "wraps around" (always False for ROOT histograms).
        """
        return False

    @property
    def discrete(self):
        """
        True if bins are discrete: if they have string-valued labels.
        """
        fNbins = self._axis.member("fNbins")
        fLabels = self._axis.member("fLabels", none_if_missing=True)
        return fLabels is not None and len(fLabels) == fNbins


class TAxis(Sequence):
    def __len__(self):
        """
        The number of bins in the axis.
        """
        return self.member("fNbins")

    def __getitem__(self, where):
        """
        Returns the label at ``where`` if it exists or the interval at ``where``.

        The indexing assumes that ``flow=False``.
        """
        fNbins = self.member("fNbins")
        fXbins = self.member("fXbins", none_if_missing=True)
        fLabels = self.member("fLabels", none_if_missing=True)

        if fLabels is not None and len(fLabels) == fNbins:
            return str(fLabels[where])

        elif fXbins is None or len(fXbins) != fNbins + 1:
            fXmin, fXmax = self.member("fXmin"), self.member("fXmax")
            low = (fXmax - fXmin) * (where) / float(fNbins) + fXmin
            high = (fXmax - fXmin) * (where + 1) / float(fNbins) + fXmin
            return low, high

        else:
            return fXbins[where], fXbins[where + 1]

    def __iter__(self):
        """
        Iterate over the output of ``__getitem__``.
        """
        fNbins = self.member("fNbins")
        fLabels = self.member("fLabels", none_if_missing=True)

        if fLabels is not None and len(fLabels) == fNbins:
            for x in fLabels:
                yield str(x)
        else:
            for low, high in self.intervals():
                yield low, high

    def __eq__(self, other):
        """
        Two axes are equal if they have the same type and
        ``list(self) == list(other)``.
        """
        if type(self) is not type(other):
            return False

        self_fNbins = self.member("fNbins")
        other_fNbins = other.member("fNbins")
        if self_fNbins != other_fNbins:
            return False

        self_fLabels = self.member("fLabels", none_if_missing=True)
        other_fLabels = other.member("fLabels", none_if_missing=True)
        self_labeled = self_fLabels is not None and len(self_fLabels) == self_fNbins
        other_labeled = other_fLabels is not None and len(other_fLabels) == other_fNbins

        if self_labeled and other_labeled:
            return all(x == y for x, y in zip(self_fLabels, other_fLabels))
        elif not self_labeled and not other_labeled:
            return numpy.array_equal(self.edges(), other.edges())
        else:
            return False

    def __ne__(self, other):
        """
        Some versions of Python don't automatically negate __eq__.
        """
        return not self.__eq__(other)

    @property
    def traits(self):
        """
        Describes read-only properties of a histogram axis.

        For example, ``axis.traits.discrete`` is True if the histogram has
        labels; False otherwise.
        """
        return AxisTraits(self)

    @property
    def low(self):
        """
        The low edge of the first normal (finite-width) bin.

        For ROOT histograms, numerical edges exist even if the axis also has
        string-valued labels.
        """
        return self.member("fXmin")

    @property
    def high(self):
        """
        The high edge of the last normal (finite-width) bin.

        For ROOT histograms, numerical edges exist even if the axis also has
        string-valued labels.
        """
        return self.member("fXmax")

    @property
    def width(self):
        """
        The average bin width (or only bin width if the binning is uniform).
        """
        fNbins = self.member("fNbins")
        fXbins = self.member("fXbins", none_if_missing=True)

        if fXbins is None or len(fXbins) != fNbins + 1:
            return (self.member("fXmax") - self.member("fXmin")) / fNbins
        else:
            return self.widths().mean()

    def labels(self, flow=False):
        """
        Args:
            flow (bool): If True, include ``"underflow"`` and ``"overflow"``
                before and after the normal (finite-width) bin labels (if they
                exist).

        If string-valued labels exist, this returns them as a Python list of
        Python strings. Otherwise, this returns None.

        Setting ``flow=True`` increases the length of the output by two.
        """
        fNbins = self.member("fNbins")
        fLabels = self.member("fLabels", none_if_missing=True)

        if fLabels is not None and len(fLabels) == fNbins:
            out = [str(x) for x in fLabels]
            if flow:
                return ["underflow"] + out + ["overflow"]
            else:
                return out
        else:
            return None

    def edges(self, flow=False):
        """
        Args:
            flow (bool): If True, include ``-inf`` and ``inf`` before and
                after the normal (finite-width) bin edges.

        Returns numerical edges between bins as a one-dimensional ``numpy.ndarray``
        of ``numpy.float64``.

        Even with ``flow=False``, the number of edges is *one greater than* the
        number of normal (finite-width) bins because they represent "fenceposts"
        between the bins, including one below and one above the full range.

        Setting ``flow=True`` increases the length of the output by two.

        For ROOT histograms, numerical edges exist even if the axis also has
        string-valued labels.
        """
        fNbins = self.member("fNbins")
        fXbins = self.member("fXbins", none_if_missing=True)

        if fXbins is None or len(fXbins) != fNbins + 1:
            fXbins = numpy.linspace(
                self.member("fXmin"), self.member("fXmax"), fNbins + 1
            )

        if flow:
            out = numpy.empty(fNbins + 3, dtype=numpy.float64)
            out[0] = -numpy.inf
            out[-1] = numpy.inf
            out[1:-1] = fXbins
        else:
            out = numpy.asarray(fXbins, dtype=fXbins.dtype.newbyteorder("="))

        return out

    def intervals(self, flow=False):
        """
        Args:
            flow (bool): If True, include ``[-inf, min]`` and ``[max, inf]``
                before and after the normal (finite-width) intervals.

        Returns low, high pairs for each bin interval as a two-dimensional
        ``numpy.ndarray`` of ``numpy.float64``.

        With ``flow=False``, the number of intervals is equal to the number of
        normal (finite-width) bins.

        Setting ``flow=True`` increases the length of the output by two.

        For ROOT histograms, numerical intervals exist even if the axis also has
        string-valued labels.
        """
        fNbins = self.member("fNbins")
        fXbins = self.member("fXbins", none_if_missing=True)

        if fXbins is None or len(fXbins) != fNbins + 1:
            fXbins = numpy.linspace(
                self.member("fXmin"), self.member("fXmax"), fNbins + 1
            )

        if flow:
            out = numpy.empty((fNbins + 2, 2), dtype=numpy.float64)
            out[0, 0] = -numpy.inf
            out[-1, 1] = numpy.inf
            out[1:, 0] = fXbins
            out[:-1, 1] = fXbins
        else:
            out = numpy.empty((fNbins, 2), dtype=numpy.float64)
            out[:, 0] = fXbins[:-1]
            out[:, 1] = fXbins[1:]

        return out

    def centers(self, flow=False):
        """
        Args:
            flow (bool): If True, include ``-inf`` and ``inf`` before and after
                the normal (finite) bin centers.

        Returns bin center positions as a one-dimensional ``numpy.ndarray`` of
        ``numpy.float64``.

        With ``flow=False``, the number of bin centers is equal to the number of
        normal (finite-width) bins.

        Setting ``flow=True`` increases the length of the output by two.

        For ROOT histograms, numerical bin centers exist even if the axis also has
        string-valued labels.
        """
        edges = self.edges(flow=flow)
        return (edges[1:] + edges[:-1]) / 2.0

    def widths(self, flow=False):
        """
        Args:
            flow (bool): If True, include ``-inf`` and ``inf`` before and after
                the normal (finite) bin widths.

        Returns bin widths as a one-dimensional ``numpy.ndarray`` of
        ``numpy.float64``.

        With ``flow=False``, the number of bin widths is equal to the number of
        normal (finite-width) bins.

        Setting ``flow=True`` increases the length of the output by two.

        For ROOT histograms, numerical bin widths exist even if the axis also has
        string-valued labels.
        """
        fNbins = self.member("fNbins")
        fXbins = self.member("fXbins", none_if_missing=True)

        if not flow and (fXbins is None or len(fXbins) != fNbins + 1):
            width = (self.member("fXmax") - self.member("fXmin")) / fNbins
            return numpy.broadcast_to(width, (fNbins,))
        else:
            edges = self.edges(flow=flow)
            return edges[1:] - edges[:-1]