manipulate.py

# -*- coding: utf-8 -*-
"""
Interactive Manipulation
"""

# This largely is not usable.
# no_doc = True


# from mathics import settings
# from mathics.core.builtin import Builtin
# from mathics.core.atoms import Integer, String
# from mathics.core.attributes import A_HOLD_ALL, A_PROTECTED
# from mathics.core.convert.python import from_python
# from mathics.core.evaluation import Output
# from mathics.core.expression import Expression
# from mathics.core.list import ListExpression
# from mathics.core.symbols import Symbol, strip_context
# from mathics.core.systemsymbols import SymbolSet

# try:
#     from ipykernel.kernelbase import Kernel

#     _jupyter = True
# except ImportError:
#     _jupyter = False

# try:
#     from IPython.core.formatters import IPythonDisplayFormatter
#     from ipywidgets import Box, DOMWidget, FloatSlider, IntSlider, ToggleButtons

#     _ipywidgets = True
# except ImportError:
#     # fallback to non-Manipulate-enabled build if we don't have ipywidgets installed.
#     _ipywidgets = False


# SymbolModule = Symbol("Module")
# SymbolReleaseHold = Symbol("ReleaseHold")

# """
# A basic implementation of Manipulate[]. There is currently no support for Dynamic[] elements.
# This implementation is basically a port from ipywidget.widgets.interaction for Mathics.
# """


# def _interactive(interact_f, kwargs_widgets):
#     # this is a modified version of interactive() in ipywidget.widgets.interaction

#     container = Box(_dom_classes=["widget-interact"])
#     container.children = [w for w in kwargs_widgets if isinstance(w, DOMWidget)]

#     def call_f(name=None, old=None, new=None):
#         kwargs = dict((widget._kwarg, widget.value) for widget in kwargs_widgets)
#         try:
#             interact_f(**kwargs)
#         except Exception as e:
#             container.log.warn("Exception in interact callback: %s", e, exc_info=True)

#     for widget in kwargs_widgets:
#         widget.on_trait_change(call_f, "value")

#     container.on_displayed(lambda _: call_f(None, None, None))

#     return container


# class IllegalWidgetArguments(Exception):
#     def __init__(self, var):
#         super(IllegalWidgetArguments, self).__init__()
#         self.var = var


# class JupyterWidgetError(Exception):
#     def __init__(self, err):
#         super(JupyterWidgetError, self).__init__()
#         self.err = err


# class ManipulateParameter(
#     Builtin
# ):  # parses one Manipulate[] parameter spec, e.g. {x, 1, 2}, see _WidgetInstantiator
#     context = "System`Private`"

#     rules = {
#         # detect x and {x, default} and {x, default, label}.
#         "System`Private`ManipulateParameter[{s_Symbol, r__}]": "System`Private`ManipulateParameter[{Symbol -> s, Label -> s}, {r}]",
#         "System`Private`ManipulateParameter[{{s_Symbol, d_}, r__}]": "System`Private`ManipulateParameter[{Symbol -> s, Default -> d, Label -> s}, {r}]",
#         "System`Private`ManipulateParameter[{{s_Symbol, d_, l_}, r__}]": "System`Private`ManipulateParameter[{Symbol -> s, Default -> d, Label -> l}, {r}]",
#         # detect different kinds of widgets. on the use of the duplicate key "Default ->", see _WidgetInstantiator.add()
#         "System`Private`ManipulateParameter[var_, {min_?RealValuedNumberQ, max_?RealValuedNumberQ}]": 'Join[{Type -> "Continuous", Minimum -> min, Maximum -> max, Default -> min}, var]',
#         "System`Private`ManipulateParameter[var_, {min_?RealValuedNumberQ, max_?RealValuedNumberQ, step_?RealValuedNumberQ}]": 'Join[{Type -> "Discrete", Minimum -> min, Maximum -> max, Step -> step, Default -> min}, var]',
#         "System`Private`ManipulateParameter[var_, {opt_List}] /; Length[opt] > 0": 'Join[{Type -> "Options", Options -> opt, Default -> Part[opt, 1]}, var]',
#     }

#     summary_text = "interactive manipulation (not implemented yet)"


# def _manipulate_label(x):  # gets the label that is displayed for a symbol or name
#     if isinstance(x, String):
#         return x.get_string_value()
#     elif isinstance(x, Symbol):
#         return strip_context(x.get_name())
#     else:
#         return str(x)


# def _create_widget(widget, **kwargs):
#     try:
#         return widget(**kwargs)
#     except Exception as e:
#         raise JupyterWidgetError(str(e))


# class _WidgetInstantiator:
#     # we do not want to have widget instances (like FloatSlider) get into the evaluation pipeline (e.g. via Expression
#     # or Atom), since there might be all kinds of problems with serialization of these widget classes.  therefore, the
#     # elegant recursive solution for parsing parameters (like in Table[]) is not feasible here; instead, we must create
#     # and use the widgets in one "transaction" here, without holding them in expressions or atoms.

#     def __init__(self):
#         self._widgets = []  # the ipywidget widgets to control the manipulated variables
#         self._parsers = (
#             {}
#         )  # lambdas to decode the widget values into Mathics expressions

#     def add(self, expression, evaluation):
#         expr = Expression("System`Private`ManipulateParameter", expression).evaluate(
#             evaluation
#         )
#         if (
#             expr.get_head_name() != "System`List"
#         ):  # if everything was parsed ok, we get a List
#             return False
#         # convert the rules given us by ManipulateParameter[] into a dict. note: duplicate keys
#         # will be overwritten, the latest one wins.
#         kwargs = {"evaluation": evaluation}
#         for rule in expr.elements:
#             if rule.get_head_name() != "System`Rule" or len(rule.elements) != 2:
#                 return False
#             kwargs[strip_context(rule.elements[0].to_python()).lower()] = rule.elements[
#                 1
#             ]
#         widget = kwargs["type"].get_string_value()
#         del kwargs["type"]
#         getattr(self, "_add_%s_widget" % widget.lower())(**kwargs)  # create the widget
#         return True

#     def get_widgets(self):
#         return self._widgets

#     def build_callback(self, callback):
#         parsers = self._parsers

#         def new_callback(**kwargs):
#             callback(
#                 **dict((name, parsers[name](value)) for (name, value) in kwargs.items())
#             )

#         return new_callback

#     def _add_continuous_widget(
#         self, symbol, label, default, minimum, maximum, evaluation
#     ):
#         minimum_value = minimum.to_python()
#         maximum_value = maximum.to_python()
#         if minimum_value > maximum_value:
#             raise IllegalWidgetArguments(symbol)
#         else:
#             defval = min(max(default.to_python(), minimum_value), maximum_value)
#             widget = _create_widget(
#                 FloatSlider, value=defval, min=minimum_value, max=maximum_value
#             )
#             self._add_widget(widget, symbol.get_name(), lambda x: from_python(x), label)

#     def _add_discrete_widget(
#         self, symbol, label, default, minimum, maximum, step, evaluation
#     ):
#         minimum_value = minimum.to_python()
#         maximum_value = maximum.to_python()
#         step_value = step.to_python()
#         if (
#             minimum_value > maximum_value
#             or step_value <= 0
#             or step_value > (maximum_value - minimum_value)
#         ):
#             raise IllegalWidgetArguments(symbol)
#         else:
#             default_value = min(max(default.to_python(), minimum_value), maximum_value)
#             if all(isinstance(x, Integer) for x in [minimum, maximum, default, step]):
#                 widget = _create_widget(
#                     IntSlider,
#                     value=default_value,
#                     min=minimum_value,
#                     max=maximum_value,
#                     step=step_value,
#                 )
#             else:
#                 widget = _create_widget(
#                     FloatSlider,
#                     value=default_value,
#                     min=minimum_value,
#                     max=maximum_value,
#                     step=step_value,
#                 )
#             self._add_widget(widget, symbol.get_name(), lambda x: from_python(x), label)

#     def _add_options_widget(self, symbol, options, default, label, evaluation):
#         formatted_options = []
#         for i, option in enumerate(options.elements):
#             data = evaluation.format_output(option, format="text")
#             formatted_options.append((data, i))

#         default_index = 0
#         for i, option in enumerate(options.elements):
#             if option.sameQ(default):
#                 default_index = i

#         widget = _create_widget(
#             ToggleButtons, options=formatted_options, value=default_index
#         )
#         self._add_widget(
#             widget, symbol.get_name(), lambda j: options.elements[j], label
#         )

#     def _add_widget(self, widget, name, parse, label):
#         if not widget.description:
#             widget.description = _manipulate_label(label)
#         widget._kwarg = name  # see _interactive() above
#         self._parsers[name] = parse
#         self._widgets.append(widget)


# class ManipulateOutput(Output):
#     def max_stored_size(self, settings):
#         return self.output.max_stored_size(settings)

#     def out(self, out):
#         return self.output.out(out)

#     def clear_output(wait=False):
#         raise NotImplementedError

#     def display_data(self, result):
#         raise NotImplementedError


# class Manipulate(Builtin):
#     """
#     <url>
#     :WMA link:
#     https://reference.wolfram.com/language/ref/Manipulate.html</url>

#     <dl>
#       <dt>'Manipulate[$expr1$, {$u$, $u_min$, $u_max$}]'
#       <dd>interactively compute and display an expression with different values of $u$.

#       <dt>'Manipulate[$expr1$, {$u$, $u_min$, $u_max$, $du$}]'
#       <dd>allows $u$ to vary between $u_min$ and $u_max$ in steps of $du$.

#       <dt>'Manipulate[$expr1$, {{$u$, $u_init$}, $u_min$, $u_max$, ...}]'
#       <dd>starts with initial value of $u_init$.

#       <dt>'Manipulate[$expr1$, {{$u$, $u_init$, $u_lbl$}, ...}]'
#       <dd>labels the $u$ controll by $u_lbl$.

#       <dt>'Manipulate[$expr1$, {$u$, {$u_1$, $u_2$, ...}}]'
#       <dd>sets $u$ to take discrete values $u_1$, $u_2$, ... .

#       <dt>'Manipulate[$expr1$, {$u$, ...}, {$v$, ...}, ...]'
#       <dd>control each of $u$, $v$, ... .
#     </dl>

#     >> Manipulate[N[Sin[y]], {y, 1, 20, 2}]
#      : Manipulate[] only works inside a Jupyter notebook.
#      = Manipulate[N[Sin[y]], {y, 1, 20, 2}]

#     >> Manipulate[i ^ 3, {i, {2, x ^ 4, a}}]
#      : Manipulate[] only works inside a Jupyter notebook.
#      = Manipulate[i ^ 3, {i, {2, x ^ 4, a}}]

#     >> Manipulate[x ^ y, {x, 1, 20}, {y, 1, 3}]
#      : Manipulate[] only works inside a Jupyter notebook.
#      = Manipulate[x ^ y, {x, 1, 20}, {y, 1, 3}]

#     >> Manipulate[N[1 / x], {{x, 1}, 0, 2}]
#      : Manipulate[] only works inside a Jupyter notebook.
#      = Manipulate[N[1 / x], {{x, 1}, 0, 2}]

#     >> Manipulate[N[1 / x], {{x, 1}, 0, 2, 0.1}]
#      : Manipulate[] only works inside a Jupyter notebook.
#      = Manipulate[N[1 / x], {{x, 1}, 0, 2, 0.1}]
#     """

#     # TODO: correct in the jupyter interface but can't be checked in tests
#     """
#     #> Manipulate[x, {x}]
#      = Manipulate[x, {x}]

#     #> Manipulate[x, {x, 1, 0}]
#      : 'Illegal variable range or step parameters for `x`.
#      = Manipulate[x, {x, 1, 0}]
#     """
#     attributes = (
#         A_HOLD_ALL | A_PROTECTED
#     )  # we'll call ReleaseHold at the time of evaluation below

#     messages = {
#         "jupyter": "Manipulate[] only works inside a Jupyter notebook.",
#         "imathics": "Your IMathics kernel does not seem to support all necessary operations. "
#         + "Please check that you have the latest version installed.",
#         "widgetmake": 'Jupyter widget construction failed with "``".',
#         "widgetargs": "Illegal variable range or step parameters for ``.",
#         "widgetdisp": "Jupyter failed to display the widget.",
#     }

#     no_doc = True  # This largely doesn't work

#     requires = ("ipywidgets",)
#     summary_text = "interactively manipulate any expression, graphic, or other object"

#     def eval(self, expr, args, evaluation):
#         "Manipulate[expr_, args__]"
#         if (not _jupyter) or (not Kernel.initialized()) or (Kernel.instance() is None):
#             evaluation.message("Manipulate", "jupyter")
#             return

#         instantiator = (
#             _WidgetInstantiator()
#         )  # knows about the arguments and their widgets

#         for arg in args.get_sequence():
#             try:
#                 if not instantiator.add(
#                     arg, evaluation
#                 ):  # not a valid argument pattern?
#                     return
#             except IllegalWidgetArguments as e:
#                 evaluation.message(
#                     "Manipulate", "widgetargs", strip_context(str(e.var))
#                 )
#                 return
#             except JupyterWidgetError as e:
#                 evaluation.message("Manipulate", "widgetmake", e.err)
#                 return
#
#         clear_output_callback = evaluation.output.clear
#         display_data_callback = evaluation.output.display  # for pushing updates

#         try:
#             clear_output_callback(wait=True)
#         except NotImplementedError:
#             evaluation.message("Manipulate", "imathics")
#             return
#         def callback(**kwargs):
#             clear_output_callback(wait=True)

#             line_no = evaluation.definitions.get_line_no()

#             vars = [
#                 Expression(SymbolSet, Symbol(name), value)
#                 for name, value in kwargs.items()
#             ]
#             evaluatable = Expression(
#                 SymbolReleaseHold, Expression(SymbolModule, ListExpression(*vars), expr)
#             )

#             result = evaluation.evaluate(evaluatable, timeout=settings.TIMEOUT)
#             if result:
#                 display_data_callback(data=result.result, metadata={})

#             evaluation.definitions.set_line_no(
#                 line_no
#             )  # do not increment line_no for manipulate computations

#         widgets = instantiator.get_widgets()
#         if len(widgets) > 0:
#             box = _interactive(
#                 instantiator.build_callback(callback), widgets
#             )  # create the widget
#             formatter = IPythonDisplayFormatter()
#             if not formatter(box):  # make the widget appear on the Jupyter notebook
#                 evaluation.message("Manipulate", "widgetdisp")
#                 return
#         return Symbol(
#             "Null"
#         )  # the interactive output is pushed via kernel.display_data_callback (see above)