lib/matplotlib/rcsetup.py

"""
The rcsetup module contains the validation code for customization using
Matplotlib's rc settings.

Each rc setting is assigned a function used to validate any attempted changes
to that setting.  The validation functions are defined in the rcsetup module,
and are used to construct the rcParams global object which stores the settings
and is referenced throughout Matplotlib.

The default values of the rc settings are set in the default matplotlibrc file.
Any additions or deletions to the parameter set listed here should also be
propagated to the :file:`lib/matplotlib/mpl-data/matplotlibrc` in Matplotlib's
root source directory.
"""

import ast
from functools import lru_cache, reduce
from numbers import Real
import operator
import os
import re

import numpy as np

import matplotlib as mpl
from matplotlib import _api, cbook
from matplotlib.backends import BackendFilter, backend_registry
from matplotlib.cbook import ls_mapper
from matplotlib.colors import Colormap, is_color_like
from matplotlib._fontconfig_pattern import parse_fontconfig_pattern
from matplotlib._enums import JoinStyle, CapStyle

# Don't let the original cycler collide with our validating cycler
from cycler import Cycler, cycler as ccycler


@_api.caching_module_getattr
class __getattr__:
    @_api.deprecated(
        "3.9",
        alternative="``matplotlib.backends.backend_registry.list_builtin"
            "(matplotlib.backends.BackendFilter.INTERACTIVE)``")
    @property
    def interactive_bk(self):
        return backend_registry.list_builtin(BackendFilter.INTERACTIVE)

    @_api.deprecated(
        "3.9",
        alternative="``matplotlib.backends.backend_registry.list_builtin"
            "(matplotlib.backends.BackendFilter.NON_INTERACTIVE)``")
    @property
    def non_interactive_bk(self):
        return backend_registry.list_builtin(BackendFilter.NON_INTERACTIVE)

    @_api.deprecated(
        "3.9",
        alternative="``matplotlib.backends.backend_registry.list_builtin()``")
    @property
    def all_backends(self):
        return backend_registry.list_builtin()


class ValidateInStrings:
    def __init__(self, key, valid, ignorecase=False, *,
                 _deprecated_since=None):
        """*valid* is a list of legal strings."""
        self.key = key
        self.ignorecase = ignorecase
        self._deprecated_since = _deprecated_since

        def func(s):
            if ignorecase:
                return s.lower()
            else:
                return s
        self.valid = {func(k): k for k in valid}

    def __call__(self, s):
        if self._deprecated_since:
            name, = (k for k, v in globals().items() if v is self)
            _api.warn_deprecated(
                self._deprecated_since, name=name, obj_type="function")
        if self.ignorecase and isinstance(s, str):
            s = s.lower()
        if s in self.valid:
            return self.valid[s]
        msg = (f"{s!r} is not a valid value for {self.key}; supported values "
               f"are {[*self.valid.values()]}")
        if (isinstance(s, str)
                and (s.startswith('"') and s.endswith('"')
                     or s.startswith("'") and s.endswith("'"))
                and s[1:-1] in self.valid):
            msg += "; remove quotes surrounding your string"
        raise ValueError(msg)


def _single_string_color_list(s, scalar_validator):
    """
    Convert the string *s* to a list of colors interpreting it either as a
    color sequence name, or a string containing single-letter colors.
    """
    try:
        colors = mpl.color_sequences[s]
    except KeyError:
        try:
            # Sometimes, a list of colors might be a single string
            # of single-letter colornames. So give that a shot.
            colors = [scalar_validator(v.strip()) for v in s if v.strip()]
        except ValueError:
            raise ValueError(f'{s!r} is neither a color sequence name nor can '
                             'it be interpreted as a list of colors')

    return colors


@lru_cache
def _listify_validator(scalar_validator, allow_stringlist=False, *,
                       n=None, doc=None):
    def f(s):
        if isinstance(s, str):
            try:
                val = [scalar_validator(v.strip()) for v in s.split(',')
                       if v.strip()]
            except Exception:
                if allow_stringlist:
                    # Special handling for colors
                    val = _single_string_color_list(s, scalar_validator)
                else:
                    raise
        # Allow any ordered sequence type -- generators, np.ndarray, pd.Series
        # -- but not sets, whose iteration order is non-deterministic.
        elif np.iterable(s) and not isinstance(s, (set, frozenset)):
            # The condition on this list comprehension will preserve the
            # behavior of filtering out any empty strings (behavior was
            # from the original validate_stringlist()), while allowing
            # any non-string/text scalar values such as numbers and arrays.
            val = [scalar_validator(v) for v in s
                   if not isinstance(v, str) or v]
        else:
            raise ValueError(
                f"Expected str or other non-set iterable, but got {s}")
        if n is not None and len(val) != n:
            raise ValueError(
                f"Expected {n} values, but there are {len(val)} values in {s}")
        return val

    try:
        f.__name__ = f"{scalar_validator.__name__}list"
    except AttributeError:  # class instance.
        f.__name__ = f"{type(scalar_validator).__name__}List"
    f.__qualname__ = f.__qualname__.rsplit(".", 1)[0] + "." + f.__name__
    f.__doc__ = doc if doc is not None else scalar_validator.__doc__
    return f


def validate_any(s):
    return s
validate_anylist = _listify_validator(validate_any)


def _validate_date(s):
    try:
        np.datetime64(s)
        return s
    except ValueError:
        raise ValueError(
            f'{s!r} should be a string that can be parsed by numpy.datetime64')


def validate_bool(b):
    """Convert b to ``bool`` or raise."""
    if isinstance(b, str):
        b = b.lower()
    if b in ('t', 'y', 'yes', 'on', 'true', '1', 1, True):
        return True
    elif b in ('f', 'n', 'no', 'off', 'false', '0', 0, False):
        return False
    else:
        raise ValueError(f'Cannot convert {b!r} to bool')


def validate_axisbelow(s):
    try:
        return validate_bool(s)
    except ValueError:
        if isinstance(s, str):
            if s == 'line':
                return 'line'
    raise ValueError(f'{s!r} cannot be interpreted as'
                     ' True, False, or "line"')


def validate_dpi(s):
    """Confirm s is string 'figure' or convert s to float or raise."""
    if s == 'figure':
        return s
    try:
        return float(s)
    except ValueError as e:
        raise ValueError(f'{s!r} is not string "figure" and '
                         f'could not convert {s!r} to float') from e


def _make_type_validator(cls, *, allow_none=False):
    """
    Return a validator that converts inputs to *cls* or raises (and possibly
    allows ``None`` as well).
    """

    def validator(s):
        if (allow_none and
                (s is None or cbook._str_lower_equal(s, "none"))):
            if cbook._str_lower_equal(s, "none") and s != "None":
                _api.warn_deprecated(
                    "3.11",
                    message=f"Using the capitalization {s!r} in matplotlibrc for "
                            "*None* is deprecated in %(removal)s and will lead to an "
                            "error from version 3.13 onward. Please use 'None' "
                            "instead."
                )
            return None
        if cls is str and not isinstance(s, str):
            raise ValueError(f'Could not convert {s!r} to str')
        try:
            return cls(s)
        except (TypeError, ValueError) as e:
            raise ValueError(
                f'Could not convert {s!r} to {cls.__name__}') from e

    validator.__name__ = f"validate_{cls.__name__}"
    if allow_none:
        validator.__name__ += "_or_None"
    validator.__qualname__ = (
        validator.__qualname__.rsplit(".", 1)[0] + "." + validator.__name__)
    return validator


validate_string = _make_type_validator(str)
validate_string_or_None = _make_type_validator(str, allow_none=True)
validate_stringlist = _listify_validator(
    validate_string, doc='return a list of strings')
validate_int = _make_type_validator(int)
validate_int_or_None = _make_type_validator(int, allow_none=True)
validate_float = _make_type_validator(float)
validate_float_or_None = _make_type_validator(float, allow_none=True)
validate_floatlist = _listify_validator(
    validate_float, doc='return a list of floats')


def _validate_marker(s):
    try:
        return validate_int(s)
    except ValueError as e:
        try:
            return validate_string(s)
        except ValueError as e:
            raise ValueError('Supported markers are [string, int]') from e


_validate_markerlist = _listify_validator(
    _validate_marker, doc='return a list of markers')


def _validate_pathlike(s):
    if isinstance(s, (str, os.PathLike)):
        # Store value as str because savefig.directory needs to distinguish
        # between "" (cwd) and "." (cwd, but gets updated by user selections).
        return os.fsdecode(s)
    else:
        return validate_string(s)


def validate_fonttype(s):
    """
    Confirm that this is a Postscript or PDF font type that we know how to
    convert to.
    """
    fonttypes = {'type3':    3,
                 'truetype': 42}
    try:
        fonttype = validate_int(s)
    except ValueError:
        try:
            return fonttypes[s.lower()]
        except KeyError as e:
            raise ValueError('Supported Postscript/PDF font types are %s'
                             % list(fonttypes)) from e
    else:
        if fonttype not in fonttypes.values():
            raise ValueError(
                'Supported Postscript/PDF font types are %s' %
                list(fonttypes.values()))
        return fonttype


_auto_backend_sentinel = object()


def validate_backend(s):
    if s is _auto_backend_sentinel or backend_registry.is_valid_backend(s):
        return s
    else:
        msg = (f"'{s}' is not a valid value for backend; supported values are "
               f"{backend_registry.list_all()}")
        raise ValueError(msg)


def _validate_toolbar(s):
    s = ValidateInStrings(
        'toolbar', ['None', 'toolbar2', 'toolmanager'], ignorecase=True)(s)
    if s == 'toolmanager':
        _api.warn_external(
            "Treat the new Tool classes introduced in v1.5 as experimental "
            "for now; the API and rcParam may change in future versions.")
    return s


def validate_color_or_inherit(s):
    """Return a valid color arg."""
    if cbook._str_equal(s, 'inherit'):
        return s
    return validate_color(s)


def validate_color_or_auto(s):
    if cbook._str_equal(s, 'auto'):
        return s
    return validate_color(s)


def _validate_color_or_edge(s):
    if cbook._str_equal(s, 'edge'):
        return s
    return validate_color(s)


def validate_color_for_prop_cycle(s):
    # N-th color cycle syntax can't go into the color cycle.
    if isinstance(s, str) and re.match("^C[0-9]$", s):
        raise ValueError(f"Cannot put cycle reference ({s!r}) in prop_cycler")
    return validate_color(s)


def _validate_color_or_linecolor(s):
    if cbook._str_equal(s, 'linecolor'):
        return s
    elif cbook._str_equal(s, 'mfc') or cbook._str_equal(s, 'markerfacecolor'):
        return 'markerfacecolor'
    elif cbook._str_equal(s, 'mec') or cbook._str_equal(s, 'markeredgecolor'):
        return 'markeredgecolor'
    elif s is None:
        return None
    elif isinstance(s, str) and len(s) == 6 or len(s) == 8:
        stmp = '#' + s
        if is_color_like(stmp):
            return stmp
        if s.lower() == 'none':
            return None
    elif is_color_like(s):
        return s

    raise ValueError(f'{s!r} does not look like a color arg')


def validate_color(s):
    """Return a valid color arg."""
    if isinstance(s, str):
        if s.lower() == 'none':
            return 'none'
        if len(s) == 6 or len(s) == 8:
            stmp = '#' + s
            if is_color_like(stmp):
                return stmp

    if is_color_like(s):
        return s

    # If it is still valid, it must be a tuple (as a string from matplotlibrc).
    try:
        color = ast.literal_eval(s)
    except (SyntaxError, ValueError):
        pass
    else:
        if is_color_like(color):
            return color

    raise ValueError(f'{s!r} does not look like a color arg')


validate_colorlist = _listify_validator(
    validate_color, allow_stringlist=True, doc='return a list of colorspecs')


def _validate_cmap(s):
    _api.check_isinstance((str, Colormap), cmap=s)
    return s


def validate_aspect(s):
    if s in ('auto', 'equal'):
        return s
    try:
        return float(s)
    except ValueError as e:
        raise ValueError('not a valid aspect specification') from e


def validate_fontsize_None(s):
    if s is None or s == 'None':
        return None
    else:
        return validate_fontsize(s)


def validate_fontsize(s):
    fontsizes = ['xx-small', 'x-small', 'small', 'medium', 'large',
                 'x-large', 'xx-large', 'smaller', 'larger']
    if isinstance(s, str):
        s = s.lower()
    if s in fontsizes:
        return s
    try:
        return float(s)
    except ValueError as e:
        raise ValueError("%s is not a valid font size. Valid font sizes "
                         "are %s." % (s, ", ".join(fontsizes))) from e


validate_fontsizelist = _listify_validator(validate_fontsize)


def validate_fontweight(s):
    weights = [
        'ultralight', 'light', 'normal', 'regular', 'book', 'medium', 'roman',
        'semibold', 'demibold', 'demi', 'bold', 'heavy', 'extra bold', 'black']
    # Note: Historically, weights have been case-sensitive in Matplotlib
    if s in weights:
        return s
    try:
        return int(s)
    except (ValueError, TypeError) as e:
        raise ValueError(f'{s} is not a valid font weight.') from e


def validate_fontstretch(s):
    stretchvalues = [
        'ultra-condensed', 'extra-condensed', 'condensed', 'semi-condensed',
        'normal', 'semi-expanded', 'expanded', 'extra-expanded',
        'ultra-expanded']
    # Note: Historically, stretchvalues have been case-sensitive in Matplotlib
    if s in stretchvalues:
        return s
    try:
        return int(s)
    except (ValueError, TypeError) as e:
        raise ValueError(f'{s} is not a valid font stretch.') from e


def validate_font_properties(s):
    parse_fontconfig_pattern(s)
    return s


def _validate_mathtext_fallback(s):
    _fallback_fonts = ['cm', 'stix', 'stixsans']
    if isinstance(s, str):
        s = s.lower()
    if s is None or s == 'none':
        return None
    elif s.lower() in _fallback_fonts:
        return s
    else:
        raise ValueError(
            f"{s} is not a valid fallback font name. Valid fallback font "
            f"names are {','.join(_fallback_fonts)}. Passing 'None' will turn "
            "fallback off.")


def validate_whiskers(s):
    try:
        return _listify_validator(validate_float, n=2)(s)
    except (TypeError, ValueError):
        try:
            return float(s)
        except ValueError as e:
            raise ValueError("Not a valid whisker value [float, "
                             "(float, float)]") from e


def validate_ps_distiller(s):
    if isinstance(s, str):
        s = s.lower()
    if s in ('none', None, 'false', False):
        return None
    else:
        return ValidateInStrings('ps.usedistiller', ['ghostscript', 'xpdf'])(s)


# A validator dedicated to the named line styles, based on the items in
# ls_mapper, and a list of possible strings read from Line2D.set_linestyle
_validate_named_linestyle = ValidateInStrings(
    'linestyle',
    [*ls_mapper.keys(), *ls_mapper.values(), 'None', 'none', ' ', ''],
    ignorecase=True)


def _validate_linestyle(ls):
    """
    A validator for all possible line styles, the named ones *and*
    the on-off ink sequences.
    """
    if isinstance(ls, str):
        try:  # Look first for a valid named line style, like '--' or 'solid'.
            return _validate_named_linestyle(ls)
        except ValueError:
            pass
        try:
            ls = ast.literal_eval(ls)  # Parsing matplotlibrc.
        except (SyntaxError, ValueError):
            pass  # Will error with the ValueError at the end.

    def _is_iterable_not_string_like(x):
        # Explicitly exclude bytes/bytearrays so that they are not
        # nonsensically interpreted as sequences of numbers (codepoints).
        return np.iterable(x) and not isinstance(x, (str, bytes, bytearray))

    if _is_iterable_not_string_like(ls):
        if len(ls) == 2 and _is_iterable_not_string_like(ls[1]):
            # (offset, (on, off, on, off, ...))
            offset, onoff = ls
        else:
            # For backcompat: (on, off, on, off, ...); the offset is implicit.
            offset = 0
            onoff = ls

        if (isinstance(offset, Real)
                and len(onoff) % 2 == 0
                and all(isinstance(elem, Real) for elem in onoff)):
            return (offset, onoff)

    raise ValueError(f"linestyle {ls!r} is not a valid on-off ink sequence.")


validate_fillstyle = ValidateInStrings(
    'markers.fillstyle', ['full', 'left', 'right', 'bottom', 'top', 'none'])


validate_fillstylelist = _listify_validator(validate_fillstyle)


def validate_markevery(s):
    """
    Validate the markevery property of a Line2D object.

    Parameters
    ----------
    s : None, int, (int, int), slice, float, (float, float), or list[int]

    Returns
    -------
    None, int, (int, int), slice, float, (float, float), or list[int]
    """
    # Validate s against type slice float int and None
    if isinstance(s, (slice, float, int, type(None))):
        return s
    # Validate s against type tuple
    if isinstance(s, tuple):
        if (len(s) == 2
                and (all(isinstance(e, int) for e in s)
                     or all(isinstance(e, float) for e in s))):
            return s
        else:
            raise TypeError(
                "'markevery' tuple must be pair of ints or of floats")
    # Validate s against type list
    if isinstance(s, list):
        if all(isinstance(e, int) for e in s):
            return s
        else:
            raise TypeError(
                "'markevery' list must have all elements of type int")
    raise TypeError("'markevery' is of an invalid type")


validate_markeverylist = _listify_validator(validate_markevery)


def validate_bbox(s):
    if isinstance(s, str):
        s = s.lower()
        if s == 'tight':
            return s
        if s == 'standard':
            return None
        raise ValueError("bbox should be 'tight' or 'standard'")
    elif s is not None:
        # Backwards compatibility. None is equivalent to 'standard'.
        raise ValueError("bbox should be 'tight' or 'standard'")
    return s


def validate_sketch(s):

    if isinstance(s, str):
        s = s.lower().strip()
        if s.startswith("(") and s.endswith(")"):
            s = s[1:-1]
    if s == 'none' or s is None:
        return None
    try:
        return tuple(_listify_validator(validate_float, n=3)(s))
    except ValueError as exc:
        raise ValueError("Expected a (scale, length, randomness) tuple") from exc


def _validate_greaterthan_minushalf(s):
    s = validate_float(s)
    if s > -0.5:
        return s
    else:
        raise RuntimeError(f'Value must be >-0.5; got {s}')


def _validate_greaterequal0_lessequal1(s):
    s = validate_float(s)
    if 0 <= s <= 1:
        return s
    else:
        raise RuntimeError(f'Value must be >=0 and <=1; got {s}')


def _validate_int_greaterequal0(s):
    s = validate_int(s)
    if s >= 0:
        return s
    else:
        raise RuntimeError(f'Value must be >=0; got {s}')


def validate_hatch(s):
    r"""
    Validate a hatch pattern.
    A hatch pattern string can have any sequence of the following
    characters: ``\ / | - + * . x o O``.
    """
    if not isinstance(s, str):
        raise ValueError("Hatch pattern must be a string")
    _api.check_isinstance(str, hatch_pattern=s)
    unknown = set(s) - {'\\', '/', '|', '-', '+', '*', '.', 'x', 'o', 'O'}
    if unknown:
        raise ValueError("Unknown hatch symbol(s): %s" % list(unknown))
    return s


validate_hatchlist = _listify_validator(validate_hatch)
validate_dashlist = _listify_validator(validate_floatlist)


def _validate_minor_tick_ndivs(n):
    """
    Validate ndiv parameter related to the minor ticks.
    It controls the number of minor ticks to be placed between
    two major ticks.
    """

    if cbook._str_lower_equal(n, 'auto'):
        return n
    try:
        n = _validate_int_greaterequal0(n)
        return n
    except (RuntimeError, ValueError):
        pass

    raise ValueError("'tick.minor.ndivs' must be 'auto' or non-negative int")


_prop_validators = {
        'color': _listify_validator(validate_color_for_prop_cycle,
                                    allow_stringlist=True),
        'linewidth': validate_floatlist,
        'linestyle': _listify_validator(_validate_linestyle),
        'facecolor': validate_colorlist,
        'edgecolor': validate_colorlist,
        'joinstyle': _listify_validator(JoinStyle),
        'capstyle': _listify_validator(CapStyle),
        'fillstyle': validate_fillstylelist,
        'markerfacecolor': validate_colorlist,
        'markersize': validate_floatlist,
        'markeredgewidth': validate_floatlist,
        'markeredgecolor': validate_colorlist,
        'markevery': validate_markeverylist,
        'alpha': validate_floatlist,
        'marker': _validate_markerlist,
        'hatch': validate_hatchlist,
        'dashes': validate_dashlist,
    }
_prop_aliases = {
        'c': 'color',
        'lw': 'linewidth',
        'ls': 'linestyle',
        'fc': 'facecolor',
        'ec': 'edgecolor',
        'mfc': 'markerfacecolor',
        'mec': 'markeredgecolor',
        'mew': 'markeredgewidth',
        'ms': 'markersize',
    }


def cycler(*args, **kwargs):
    """
    Create a `~cycler.Cycler` object much like :func:`cycler.cycler`,
    but includes input validation.

    Call signatures::

      cycler(cycler)
      cycler(label=values, label2=values2, ...)
      cycler(label, values)

    Form 1 copies a given `~cycler.Cycler` object.

    Form 2 creates a `~cycler.Cycler` which cycles over one or more
    properties simultaneously. If multiple properties are given, their
    value lists must have the same length.

    Form 3 creates a `~cycler.Cycler` for a single property. This form
    exists for compatibility with the original cycler. Its use is
    discouraged in favor of the kwarg form, i.e. ``cycler(label=values)``.

    Parameters
    ----------
    cycler : Cycler
        Copy constructor for Cycler.

    label : str
        The property key. Must be a valid `.Artist` property.
        For example, 'color' or 'linestyle'. Aliases are allowed,
        such as 'c' for 'color' and 'lw' for 'linewidth'.

    values : iterable
        Finite-length iterable of the property values. These values
        are validated and will raise a ValueError if invalid.

    Returns
    -------
    Cycler
        A new :class:`~cycler.Cycler` for the given properties.

    Examples
    --------
    Creating a cycler for a single property:

    >>> c = cycler(color=['red', 'green', 'blue'])

    Creating a cycler for simultaneously cycling over multiple properties
    (e.g. red circle, green plus, blue cross):

    >>> c = cycler(color=['red', 'green', 'blue'],
    ...            marker=['o', '+', 'x'])

    """
    if args and kwargs:
        raise TypeError("cycler() can only accept positional OR keyword "
                        "arguments -- not both.")
    elif not args and not kwargs:
        raise TypeError("cycler() must have positional OR keyword arguments")

    if len(args) == 1:
        if not isinstance(args[0], Cycler):
            raise TypeError("If only one positional argument given, it must "
                            "be a Cycler instance.")
        return validate_cycler(args[0])
    elif len(args) == 2:
        pairs = [(args[0], args[1])]
    elif len(args) > 2:
        raise _api.nargs_error('cycler', '0-2', len(args))
    else:
        pairs = kwargs.items()

    validated = []
    for prop, vals in pairs:
        norm_prop = _prop_aliases.get(prop, prop)
        validator = _prop_validators.get(norm_prop, None)
        if validator is None:
            raise TypeError("Unknown artist property: %s" % prop)
        vals = validator(vals)
        # We will normalize the property names as well to reduce
        # the amount of alias handling code elsewhere.
        validated.append((norm_prop, vals))

    return reduce(operator.add, (ccycler(k, v) for k, v in validated))


class _DunderChecker(ast.NodeVisitor):
    def visit_Attribute(self, node):
        if node.attr.startswith("__") and node.attr.endswith("__"):
            raise ValueError("cycler strings with dunders are forbidden")
        self.generic_visit(node)


# A validator dedicated to the named legend loc
_validate_named_legend_loc = ValidateInStrings(
    'legend.loc',
    [
        "best",
        "upper right", "upper left", "lower left", "lower right", "right",
        "center left", "center right", "lower center", "upper center",
        "center"],
    ignorecase=True)


def _validate_legend_loc(loc):
    """
    Confirm that loc is a type which rc.Params["legend.loc"] supports.

    .. versionadded:: 3.8

    Parameters
    ----------
    loc : str | int | (float, float) | str((float, float))
        The location of the legend.

    Returns
    -------
    loc : str | int | (float, float) or raise ValueError exception
        The location of the legend.
    """
    if isinstance(loc, str):
        try:
            return _validate_named_legend_loc(loc)
        except ValueError:
            pass
        try:
            loc = ast.literal_eval(loc)
        except (SyntaxError, ValueError):
            pass
    if isinstance(loc, int):
        if 0 <= loc <= 10:
            return loc
    if isinstance(loc, tuple):
        if len(loc) == 2 and all(isinstance(e, Real) for e in loc):
            return loc
    raise ValueError(f"{loc} is not a valid legend location.")


def validate_cycler(s):
    """Return a Cycler object from a string repr or the object itself."""
    if isinstance(s, str):
        # TODO: We might want to rethink this...
        # While I think I have it quite locked down, it is execution of
        # arbitrary code without sanitation.
        # Combine this with the possibility that rcparams might come from the
        # internet (future plans), this could be downright dangerous.
        # I locked it down by only having the 'cycler()' function available.
        # UPDATE: Partly plugging a security hole.
        # I really should have read this:
        # https://nedbatchelder.com/blog/201206/eval_really_is_dangerous.html
        # We should replace this eval with a combo of PyParsing and
        # ast.literal_eval()
        try:
            _DunderChecker().visit(ast.parse(s))
            s = eval(s, {'cycler': cycler, '__builtins__': {}})
        except BaseException as e:
            raise ValueError(f"{s!r} is not a valid cycler construction: {e}"
                             ) from e
    # Should make sure what comes from the above eval()
    # is a Cycler object.
    if isinstance(s, Cycler):
        cycler_inst = s
    else:
        raise ValueError(f"Object is not a string or Cycler instance: {s!r}")

    unknowns = cycler_inst.keys - (set(_prop_validators) | set(_prop_aliases))
    if unknowns:
        raise ValueError("Unknown artist properties: %s" % unknowns)

    # Not a full validation, but it'll at least normalize property names
    # A fuller validation would require v0.10 of cycler.
    checker = set()
    for prop in cycler_inst.keys:
        norm_prop = _prop_aliases.get(prop, prop)
        if norm_prop != prop and norm_prop in cycler_inst.keys:
            raise ValueError(f"Cannot specify both {norm_prop!r} and alias "
                             f"{prop!r} in the same prop_cycle")
        if norm_prop in checker:
            raise ValueError(f"Another property was already aliased to "
                             f"{norm_prop!r}. Collision normalizing {prop!r}.")
        checker.update([norm_prop])

    # This is just an extra-careful check, just in case there is some
    # edge-case I haven't thought of.
    assert len(checker) == len(cycler_inst.keys)

    # Now, it should be safe to mutate this cycler
    for prop in cycler_inst.keys:
        norm_prop = _prop_aliases.get(prop, prop)
        cycler_inst.change_key(prop, norm_prop)

    for key, vals in cycler_inst.by_key().items():
        _prop_validators[key](vals)

    return cycler_inst


def validate_hist_bins(s):
    valid_strs = ["auto", "sturges", "fd", "doane", "scott", "rice", "sqrt"]
    if isinstance(s, str) and s in valid_strs:
        return s
    try:
        return int(s)
    except (TypeError, ValueError):
        pass
    try:
        return validate_floatlist(s)
    except ValueError:
        pass
    raise ValueError(f"'hist.bins' must be one of {valid_strs}, an int or"
                     " a sequence of floats")


class _ignorecase(list):
    """A marker class indicating that a list-of-str is case-insensitive."""


def _convert_validator_spec(key, conv):
    if isinstance(conv, list):
        ignorecase = isinstance(conv, _ignorecase)
        return ValidateInStrings(key, conv, ignorecase=ignorecase)
    else:
        return conv


# Mapping of rcParams to validators.
# Converters given as lists or _ignorecase are converted to ValidateInStrings
# immediately below.
# The rcParams defaults are defined in lib/matplotlib/mpl-data/matplotlibrc, which
# gets copied to matplotlib/mpl-data/matplotlibrc by the setup script.
_validators = {
    "backend":           validate_backend,
    "backend_fallback":  validate_bool,
    "figure.hooks":      validate_stringlist,
    "toolbar":           _validate_toolbar,
    "interactive":       validate_bool,
    "timezone":          validate_string,

    "webagg.port":            validate_int,
    "webagg.address":         validate_string,
    "webagg.open_in_browser": validate_bool,
    "webagg.port_retries":    validate_int,

    # line props
    "lines.linewidth":       validate_float,  # line width in points
    "lines.linestyle":       _validate_linestyle,  # solid line
    "lines.color":           validate_color,  # first color in color cycle
    "lines.marker":          _validate_marker,  # marker name
    "lines.markerfacecolor": validate_color_or_auto,  # default color
    "lines.markeredgecolor": validate_color_or_auto,  # default color
    "lines.markeredgewidth": validate_float,
    "lines.markersize":      validate_float,  # markersize, in points
    "lines.antialiased":     validate_bool,  # antialiased (no jaggies)
    "lines.dash_joinstyle":  JoinStyle,
    "lines.solid_joinstyle": JoinStyle,
    "lines.dash_capstyle":   CapStyle,
    "lines.solid_capstyle":  CapStyle,
    "lines.dashed_pattern":  validate_floatlist,
    "lines.dashdot_pattern": validate_floatlist,
    "lines.dotted_pattern":  validate_floatlist,
    "lines.scale_dashes":    validate_bool,

    # marker props
    "markers.fillstyle": validate_fillstyle,

    ## pcolor(mesh) props:
    "pcolor.shading": ["auto", "flat", "nearest", "gouraud"],
    "pcolormesh.snap": validate_bool,

    ## patch props
    "patch.linewidth":       validate_float,  # line width in points
    "patch.edgecolor":       validate_color,
    "patch.force_edgecolor": validate_bool,
    "patch.facecolor":       validate_color,  # first color in cycle
    "patch.antialiased":     validate_bool,  # antialiased (no jaggies)

    ## hatch props
    "hatch.color":     _validate_color_or_edge,
    "hatch.linewidth": validate_float,

    ## Histogram properties
    "hist.bins": validate_hist_bins,

    ## Boxplot properties
    "boxplot.notch":       validate_bool,
    "boxplot.vertical":    validate_bool,
    "boxplot.whiskers":    validate_whiskers,
    "boxplot.bootstrap":   validate_int_or_None,
    "boxplot.patchartist": validate_bool,
    "boxplot.showmeans":   validate_bool,
    "boxplot.showcaps":    validate_bool,
    "boxplot.showbox":     validate_bool,
    "boxplot.showfliers":  validate_bool,
    "boxplot.meanline":    validate_bool,

    "boxplot.flierprops.color":           validate_color,
    "boxplot.flierprops.marker":          _validate_marker,
    "boxplot.flierprops.markerfacecolor": validate_color_or_auto,
    "boxplot.flierprops.markeredgecolor": validate_color,
    "boxplot.flierprops.markeredgewidth": validate_float,
    "boxplot.flierprops.markersize":      validate_float,
    "boxplot.flierprops.linestyle":       _validate_linestyle,
    "boxplot.flierprops.linewidth":       validate_float,

    "boxplot.boxprops.color":     validate_color,
    "boxplot.boxprops.linewidth": validate_float,
    "boxplot.boxprops.linestyle": _validate_linestyle,

    "boxplot.whiskerprops.color":     validate_color,
    "boxplot.whiskerprops.linewidth": validate_float,
    "boxplot.whiskerprops.linestyle": _validate_linestyle,

    "boxplot.capprops.color":     validate_color,
    "boxplot.capprops.linewidth": validate_float,
    "boxplot.capprops.linestyle": _validate_linestyle,

    "boxplot.medianprops.color":     validate_color,
    "boxplot.medianprops.linewidth": validate_float,
    "boxplot.medianprops.linestyle": _validate_linestyle,

    "boxplot.meanprops.color":           validate_color,
    "boxplot.meanprops.marker":          _validate_marker,
    "boxplot.meanprops.markerfacecolor": validate_color,
    "boxplot.meanprops.markeredgecolor": validate_color,
    "boxplot.meanprops.markersize":      validate_float,
    "boxplot.meanprops.linestyle":       _validate_linestyle,
    "boxplot.meanprops.linewidth":       validate_float,

    ## font props
    "font.enable_last_resort":     validate_bool,
    "font.family":     validate_stringlist,  # used by text object
    "font.style":      validate_string,
    "font.variant":    validate_string,
    "font.stretch":    validate_fontstretch,
    "font.weight":     validate_fontweight,
    "font.size":       validate_float,  # Base font size in points
    "font.serif":      validate_stringlist,
    "font.sans-serif": validate_stringlist,
    "font.cursive":    validate_stringlist,
    "font.fantasy":    validate_stringlist,
    "font.monospace":  validate_stringlist,

    # text props
    "text.color":          validate_color,
    "text.usetex":         validate_bool,
    "text.latex.preamble": validate_string,
    "text.hinting":        ["default", "no_autohint", "force_autohint",
                            "no_hinting", "auto", "native", "either", "none"],
    "text.hinting_factor": validate_int,
    "text.kerning_factor": validate_int,
    "text.antialiased":    validate_bool,
    "text.parse_math":     validate_bool,

    "mathtext.cal":            validate_font_properties,
    "mathtext.rm":             validate_font_properties,
    "mathtext.tt":             validate_font_properties,
    "mathtext.it":             validate_font_properties,
    "mathtext.bf":             validate_font_properties,
    "mathtext.bfit":           validate_font_properties,
    "mathtext.sf":             validate_font_properties,
    "mathtext.fontset":        ["dejavusans", "dejavuserif", "cm", "stix",
                                "stixsans", "custom"],
    "mathtext.default":        ["rm", "cal", "bfit", "it", "tt", "sf", "bf", "default",
                                "bb", "frak", "scr", "regular"],
    "mathtext.fallback":       _validate_mathtext_fallback,

    "image.aspect":              validate_aspect,  # equal, auto, a number
    "image.interpolation":       validate_string,
    "image.interpolation_stage": ["auto", "data", "rgba"],
    "image.cmap":                _validate_cmap,  # gray, jet, etc.
    "image.lut":                 validate_int,  # lookup table
    "image.origin":              ["upper", "lower"],
    "image.resample":            validate_bool,
    # Specify whether vector graphics backends will combine all images on a
    # set of Axes into a single composite image
    "image.composite_image": validate_bool,

    # contour props
    "contour.negative_linestyle": _validate_linestyle,
    "contour.corner_mask":        validate_bool,
    "contour.linewidth":          validate_float_or_None,
    "contour.algorithm":          ["mpl2005", "mpl2014", "serial", "threaded"],

    # errorbar props
    "errorbar.capsize": validate_float,

    # axis props
    # alignment of x/y axis title
    "xaxis.labellocation": ["left", "center", "right"],
    "yaxis.labellocation": ["bottom", "center", "top"],

    # Axes props
    "axes.axisbelow":        validate_axisbelow,
    "axes.facecolor":        validate_color,  # background color
    "axes.edgecolor":        validate_color,  # edge color
    "axes.linewidth":        validate_float,  # edge linewidth

    "axes.spines.left":      validate_bool,  # Set visibility of axes spines,
    "axes.spines.right":     validate_bool,  # i.e., the lines around the chart
    "axes.spines.bottom":    validate_bool,  # denoting data boundary.
    "axes.spines.top":       validate_bool,

    "axes.titlesize":     validate_fontsize,  # Axes title fontsize
    "axes.titlelocation": ["left", "center", "right"],  # Axes title alignment
    "axes.titleweight":   validate_fontweight,  # Axes title font weight
    "axes.titlecolor":    validate_color_or_auto,  # Axes title font color
    # title location, axes units, None means auto
    "axes.titley":        validate_float_or_None,
    # pad from Axes top decoration to title in points
    "axes.titlepad":      validate_float,
    "axes.grid":          validate_bool,  # display grid or not
    "axes.grid.which":    ["minor", "both", "major"],  # which grids are drawn
    "axes.grid.axis":     ["x", "y", "both"],  # grid type
    "axes.labelsize":     validate_fontsize,  # fontsize of x & y labels
    "axes.labelpad":      validate_float,  # space between label and axis
    "axes.labelweight":   validate_fontweight,  # fontsize of x & y labels
    "axes.labelcolor":    validate_color,  # color of axis label
    # use scientific notation if log10 of the axis range is smaller than the
    # first or larger than the second
    "axes.formatter.limits": _listify_validator(validate_int, n=2),
    # use current locale to format ticks
    "axes.formatter.use_locale": validate_bool,
    "axes.formatter.use_mathtext": validate_bool,
    # minimum exponent to format in scientific notation
    "axes.formatter.min_exponent": validate_int,
    "axes.formatter.useoffset": validate_bool,
    "axes.formatter.offset_threshold": validate_int,
    "axes.unicode_minus": validate_bool,
    # This entry can be either a cycler object or a string repr of a
    # cycler-object, which gets eval()'ed to create the object.
    "axes.prop_cycle": validate_cycler,
    # If "data", axes limits are set close to the data.
    # If "round_numbers" axes limits are set to the nearest round numbers.
    "axes.autolimit_mode": ["data", "round_numbers"],
    "axes.xmargin": _validate_greaterthan_minushalf,  # margin added to xaxis
    "axes.ymargin": _validate_greaterthan_minushalf,  # margin added to yaxis
    "axes.zmargin": _validate_greaterthan_minushalf,  # margin added to zaxis

    "polaraxes.grid":    validate_bool,  # display polar grid or not
    "axes3d.grid":       validate_bool,  # display 3d grid
    "axes3d.automargin": validate_bool,  # automatically add margin when
                                         # manually setting 3D axis limits

    "axes3d.xaxis.panecolor":    validate_color,  # 3d background pane
    "axes3d.yaxis.panecolor":    validate_color,  # 3d background pane
    "axes3d.zaxis.panecolor":    validate_color,  # 3d background pane

    "axes3d.depthshade": validate_bool,  # depth shade for 3D scatter plots
    "axes3d.depthshade_minalpha": validate_float,  # min alpha value for depth shading

    "axes3d.mouserotationstyle": ["azel", "trackball", "sphere", "arcball"],
    "axes3d.trackballsize": validate_float,
    "axes3d.trackballborder": validate_float,

    # scatter props
    "scatter.marker":     _validate_marker,
    "scatter.edgecolors": validate_string,

    "date.epoch": _validate_date,
    "date.autoformatter.year":        validate_string,
    "date.autoformatter.month":       validate_string,
    "date.autoformatter.day":         validate_string,
    "date.autoformatter.hour":        validate_string,
    "date.autoformatter.minute":      validate_string,
    "date.autoformatter.second":      validate_string,
    "date.autoformatter.microsecond": validate_string,

    'date.converter':          ['auto', 'concise'],
    # for auto date locator, choose interval_multiples
    'date.interval_multiples': validate_bool,

    # legend properties
    "legend.fancybox": validate_bool,
    "legend.loc": _validate_legend_loc,

    # the number of points in the legend line
    "legend.numpoints":      validate_int,
    # the number of points in the legend line for scatter
    "legend.scatterpoints":  validate_int,
    "legend.fontsize":       validate_fontsize,
    "legend.title_fontsize": validate_fontsize_None,
    # color of the legend
    "legend.labelcolor":     _validate_color_or_linecolor,
    # the relative size of legend markers vs. original
    "legend.markerscale":    validate_float,
    # using dict in rcParams not yet supported, so make sure it is bool
    "legend.shadow":         validate_bool,
    # whether or not to draw a frame around legend
    "legend.frameon":        validate_bool,
    # alpha value of the legend frame
    "legend.framealpha":     validate_float_or_None,

    ## the following dimensions are in fraction of the font size
    "legend.borderpad":      validate_float,  # units are fontsize
    # the vertical space between the legend entries
    "legend.labelspacing":   validate_float,
    # the length of the legend lines
    "legend.handlelength":   validate_float,
    # the length of the legend lines
    "legend.handleheight":   validate_float,
    # the space between the legend line and legend text
    "legend.handletextpad":  validate_float,
    # the border between the Axes and legend edge
    "legend.borderaxespad":  validate_float,
    # the border between the Axes and legend edge
    "legend.columnspacing":  validate_float,
    "legend.facecolor":      validate_color_or_inherit,
    "legend.edgecolor":      validate_color_or_inherit,

    # tick properties
    "xtick.top":           validate_bool,      # draw ticks on top side
    "xtick.bottom":        validate_bool,      # draw ticks on bottom side
    "xtick.labeltop":      validate_bool,      # draw label on top
    "xtick.labelbottom":   validate_bool,      # draw label on bottom
    "xtick.major.size":    validate_float,     # major xtick size in points
    "xtick.minor.size":    validate_float,     # minor xtick size in points
    "xtick.major.width":   validate_float,     # major xtick width in points
    "xtick.minor.width":   validate_float,     # minor xtick width in points
    "xtick.major.pad":     validate_float,     # distance to label in points
    "xtick.minor.pad":     validate_float,     # distance to label in points
    "xtick.color":         validate_color,     # color of xticks
    "xtick.labelcolor":    validate_color_or_inherit,  # color of xtick labels
    "xtick.minor.visible": validate_bool,      # visibility of minor xticks
    "xtick.minor.top":     validate_bool,      # draw top minor xticks
    "xtick.minor.bottom":  validate_bool,      # draw bottom minor xticks
    "xtick.major.top":     validate_bool,      # draw top major xticks
    "xtick.major.bottom":  validate_bool,      # draw bottom major xticks
    # number of minor xticks
    "xtick.minor.ndivs":   _validate_minor_tick_ndivs,
    "xtick.labelsize":     validate_fontsize,  # fontsize of xtick labels
    "xtick.direction":     ["out", "in", "inout"],  # direction of xticks
    "xtick.alignment":     ["center", "right", "left"],

    "ytick.left":          validate_bool,      # draw ticks on left side
    "ytick.right":         validate_bool,      # draw ticks on right side
    "ytick.labelleft":     validate_bool,      # draw tick labels on left side
    "ytick.labelright":    validate_bool,      # draw tick labels on right side
    "ytick.major.size":    validate_float,     # major ytick size in points
    "ytick.minor.size":    validate_float,     # minor ytick size in points
    "ytick.major.width":   validate_float,     # major ytick width in points
    "ytick.minor.width":   validate_float,     # minor ytick width in points
    "ytick.major.pad":     validate_float,     # distance to label in points
    "ytick.minor.pad":     validate_float,     # distance to label in points
    "ytick.color":         validate_color,     # color of yticks
    "ytick.labelcolor":    validate_color_or_inherit,  # color of ytick labels
    "ytick.minor.visible": validate_bool,      # visibility of minor yticks
    "ytick.minor.left":    validate_bool,      # draw left minor yticks
    "ytick.minor.right":   validate_bool,      # draw right minor yticks
    "ytick.major.left":    validate_bool,      # draw left major yticks
    "ytick.major.right":   validate_bool,      # draw right major yticks
    # number of minor yticks
    "ytick.minor.ndivs":   _validate_minor_tick_ndivs,
    "ytick.labelsize":     validate_fontsize,  # fontsize of ytick labels
    "ytick.direction":     ["out", "in", "inout"],  # direction of yticks
    "ytick.alignment":     [
        "center", "top", "bottom", "baseline", "center_baseline"],

    "grid.color":        validate_color,  # grid color
    "grid.linestyle":    _validate_linestyle,  # solid
    "grid.linewidth":    validate_float,     # in points
    "grid.alpha":        validate_float,

    ## figure props
    # figure title
    "figure.titlesize":   validate_fontsize,
    "figure.titleweight": validate_fontweight,

    # figure labels
    "figure.labelsize":   validate_fontsize,
    "figure.labelweight": validate_fontweight,

    # figure size in inches: width by height
    "figure.figsize":          _listify_validator(validate_float, n=2),
    "figure.dpi":              validate_float,
    "figure.facecolor":        validate_color,
    "figure.edgecolor":        validate_color,
    "figure.frameon":          validate_bool,
    "figure.autolayout":       validate_bool,
    "figure.max_open_warning": validate_int,
    "figure.raise_window":     validate_bool,
    "macosx.window_mode":      ["system", "tab", "window"],

    "figure.subplot.left":   validate_float,
    "figure.subplot.right":  validate_float,
    "figure.subplot.bottom": validate_float,
    "figure.subplot.top":    validate_float,
    "figure.subplot.wspace": validate_float,
    "figure.subplot.hspace": validate_float,

    "figure.constrained_layout.use": validate_bool,  # run constrained_layout?
    # wspace and hspace are fraction of adjacent subplots to use for space.
    # Much smaller than above because we don't need room for the text.
    "figure.constrained_layout.hspace": validate_float,
    "figure.constrained_layout.wspace": validate_float,
    # buffer around the Axes, in inches.
    "figure.constrained_layout.h_pad": validate_float,
    "figure.constrained_layout.w_pad": validate_float,

    ## Saving figure's properties
    'savefig.dpi':          validate_dpi,
    'savefig.facecolor':    validate_color_or_auto,
    'savefig.edgecolor':    validate_color_or_auto,
    'savefig.orientation':  ['landscape', 'portrait'],
    "savefig.format":       validate_string,
    "savefig.bbox":         validate_bbox,  # "tight", or "standard" (= None)
    "savefig.pad_inches":   validate_float,
    # default directory in savefig dialog box
    "savefig.directory":    _validate_pathlike,
    "savefig.transparent":  validate_bool,

    "tk.window_focus": validate_bool,  # Maintain shell focus for TkAgg

    # Set the papersize/type
    "ps.papersize":       _ignorecase(
                                ["figure", "letter", "legal", "ledger",
                                 *[f"{ab}{i}" for ab in "ab" for i in range(11)]]),
    "ps.useafm":          validate_bool,
    # use ghostscript or xpdf to distill ps output
    "ps.usedistiller":    validate_ps_distiller,
    "ps.distiller.res":   validate_int,  # dpi
    "ps.fonttype":        validate_fonttype,  # 3 (Type3) or 42 (Truetype)
    "pdf.compression":    validate_int,  # 0-9 compression level; 0 to disable
    "pdf.inheritcolor":   validate_bool,  # skip color setting commands
    # use only the 14 PDF core fonts embedded in every PDF viewing application
    "pdf.use14corefonts": validate_bool,
    "pdf.fonttype":       validate_fonttype,  # 3 (Type3) or 42 (Truetype)

    "pgf.texsystem": ["xelatex", "lualatex", "pdflatex"],  # latex variant used
    "pgf.rcfonts":   validate_bool,  # use mpl's rc settings for font config
    "pgf.preamble":  validate_string,  # custom LaTeX preamble

    # write raster image data into the svg file
    "svg.image_inline": validate_bool,
    "svg.fonttype": ["none", "path"],  # save text as text ("none") or "paths"
    "svg.hashsalt": validate_string_or_None,
    "svg.id": validate_string_or_None,

    # set this when you want to generate hardcopy docstring
    "docstring.hardcopy": validate_bool,

    "path.simplify":           validate_bool,
    "path.simplify_threshold": _validate_greaterequal0_lessequal1,
    "path.snap":               validate_bool,
    "path.sketch":             validate_sketch,
    "path.effects":            validate_anylist,
    "agg.path.chunksize":      validate_int,  # 0 to disable chunking

    # key-mappings (multi-character mappings should be a list/tuple)
    "keymap.fullscreen": validate_stringlist,
    "keymap.home":       validate_stringlist,
    "keymap.back":       validate_stringlist,
    "keymap.forward":    validate_stringlist,
    "keymap.pan":        validate_stringlist,
    "keymap.zoom":       validate_stringlist,
    "keymap.save":       validate_stringlist,
    "keymap.quit":       validate_stringlist,
    "keymap.quit_all":   validate_stringlist,  # e.g.: "W", "cmd+W", "Q"
    "keymap.grid":       validate_stringlist,
    "keymap.grid_minor": validate_stringlist,
    "keymap.yscale":     validate_stringlist,
    "keymap.xscale":     validate_stringlist,
    "keymap.help":       validate_stringlist,
    "keymap.copy":       validate_stringlist,

    # Animation settings
    "animation.html":         ["html5", "jshtml", "none"],
    # Limit, in MB, of size of base64 encoded animation in HTML
    # (i.e. IPython notebook)
    "animation.embed_limit":  validate_float,
    "animation.writer":       validate_string,
    "animation.codec":        validate_string,
    "animation.bitrate":      validate_int,
    # Controls image format when frames are written to disk
    "animation.frame_format": ["png", "jpeg", "tiff", "raw", "rgba", "ppm",
                               "sgi", "bmp", "pbm", "svg"],
    # Path to ffmpeg binary. If just binary name, subprocess uses $PATH.
    "animation.ffmpeg_path":  _validate_pathlike,
    # Additional arguments for ffmpeg movie writer (using pipes)
    "animation.ffmpeg_args":  validate_stringlist,
     # Path to convert binary. If just binary name, subprocess uses $PATH.
    "animation.convert_path": _validate_pathlike,
     # Additional arguments for convert movie writer (using pipes)
    "animation.convert_args": validate_stringlist,

    # Classic (pre 2.0) compatibility mode
    # This is used for things that are hard to make backward compatible
    # with a sane rcParam alone.  This does *not* turn on classic mode
    # altogether.  For that use `matplotlib.style.use("classic")`.
    "_internal.classic_mode": validate_bool
}
_hardcoded_defaults = {  # Defaults not inferred from
    # lib/matplotlib/mpl-data/matplotlibrc...
    # ... because they are private:
    "_internal.classic_mode": False,
    # ... because they are deprecated:
    # No current deprecations.
    # backend is handled separately when constructing rcParamsDefault.
}
_validators = {k: _convert_validator_spec(k, conv)
               for k, conv in _validators.items()}