figure_resampler_interface

Abstract AbstractFigureAggregator interface for the concrete Resampler classes.

AbstractFigureAggregator

Bases: BaseFigure, ABC

Abstract interface for data aggregation functionality for plotly figures.

Source code in plotly_resampler/figure_resampler/figure_resampler_interface.py

class AbstractFigureAggregator(BaseFigure, ABC):
    """Abstract interface for data aggregation functionality for plotly figures."""

    _high_frequency_traces = ["scatter", "scattergl"]

    def __init__(
        self,
        figure: BaseFigure,
        convert_existing_traces: bool = True,
        default_n_shown_samples: int = 1000,
        default_downsampler: AbstractAggregator = MinMaxLTTB(),
        default_gap_handler: AbstractGapHandler = MedDiffGapHandler(),
        resampled_trace_prefix_suffix: Tuple[str, str] = (
            '<b style="color:sandybrown">[R]</b> ',
            "",
        ),
        show_mean_aggregation_size: bool = True,
        convert_traces_kwargs: dict | None = None,
        verbose: bool = False,
    ):
        """Instantiate a resampling data mirror.

        Parameters
        ----------
        figure: BaseFigure
            The figure that will be decorated. Can be either an empty figure
            (e.g., ``go.Figure()``, ``make_subplots()``, ``go.FigureWidget``) or an
            existing figure.
        convert_existing_traces: bool
            A bool indicating whether the high-frequency traces of the passed ``figure``
            should be resampled, by default True. Hence, when set to False, the
            high-frequency traces of the passed ``figure`` will not be resampled.
        default_n_shown_samples: int, optional
            The default number of samples that will be shown for each trace,
            by default 1000.\n
            !!! note
                * This can be overridden within the [`add_trace`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_trace] method.
                * If a trace withholds fewer datapoints than this parameter,
                  the data will *not* be aggregated.
        default_downsampler: AbstractAggregator
            An instance which implements the AbstractSeriesDownsampler interface and
            will be used as default downsampler, by default ``MinMaxLTTB``. \n
            !!! note
                This can be overridden within the [`add_trace`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_trace] method.
        default_gap_handler: GapHandler
            An instance which implements the AbstractGapHandler interface and will be
            used as default gap handler, by default ``MedDiffGapHandler``. \n
            !!! note
                This can be overridden within the [`add_trace`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_trace] method.
        resampled_trace_prefix_suffix: str, optional
            A tuple which contains the ``prefix`` and ``suffix``, respectively, which
            will be added to the trace its legend-name when a resampled version of the
            trace is shown. By default, a bold, orange ``[R]`` is shown as prefix
            (no suffix is shown).
        show_mean_aggregation_size: bool, optional
            Whether the mean aggregation bin size will be added as a suffix to the trace
            its legend-name, by default True.
        convert_traces_kwargs: dict, optional
            A dict of kwargs that will be passed to the [`add_traces`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_traces] method and
            will be used to convert the existing traces. \n
            !!! note
                This argument is only used when the passed ``figure`` contains data and
                ``convert_existing_traces`` is set to True.
        verbose: bool, optional
            Whether some verbose messages will be printed or not, by default False.

        """
        self._hf_data: Dict[str, dict] = {}
        self._global_n_shown_samples = default_n_shown_samples
        self._print_verbose = verbose
        self._show_mean_aggregation_size = show_mean_aggregation_size

        assert len(resampled_trace_prefix_suffix) == 2
        self._prefix, self._suffix = resampled_trace_prefix_suffix

        self._global_downsampler = default_downsampler
        self._global_gap_handler = default_gap_handler

        # Given figure should always be a BaseFigure that is not wrapped by
        # a plotly-resampler class
        assert isinstance(figure, BaseFigure)
        assert not issubclass(type(figure), AbstractFigureAggregator)
        self._figure_class = figure.__class__

        # Overwrite the passed arguments with the property dict values
        # (this is the case when the PR figure is created from a pickled object)
        if hasattr(figure, "_pr_props"):
            pr_props = figure._pr_props  # a dict of PR properties
            if pr_props is not None:
                # Overwrite the default arguments with the serialized properties
                for k, v in pr_props.items():
                    setattr(self, k, v)
            delattr(figure, "_pr_props")  # should not be stored anymore

        if convert_existing_traces:
            # call __init__ with the correct layout and set the `_grid_ref` of the
            # to-be-converted figure
            f_ = self._figure_class(layout=figure.layout)
            f_._grid_str = figure._grid_str
            f_._grid_ref = figure._grid_ref
            super().__init__(f_)

            if convert_traces_kwargs is None:
                convert_traces_kwargs = {}

            # make sure that the UIDs of these traces do not get adjusted
            self._data_validator.set_uid = False
            self.add_traces(figure.data, **convert_traces_kwargs)
        else:
            super().__init__(figure)
            self._data_validator.set_uid = False

        # A list of al xaxis and yaxis string names
        # e.g., "xaxis", "xaxis2", "xaxis3", .... for _xaxis_list
        self._xaxis_list = self._re_matches(re.compile("xaxis\d*"), self._layout.keys())
        self._yaxis_list = self._re_matches(re.compile("yaxis\d*"), self._layout.keys())
        # edge case: an empty `go.Figure()` does not yet contain axes keys
        if not len(self._xaxis_list):
            self._xaxis_list = ["xaxis"]
            self._yaxis_list = ["yaxis"]

        # Make sure to reset the layout its range
        # self.update_layout(
        #     {
        #         axis: {"autorange": None, "range": None}
        #         for axis in self._xaxis_list + self._yaxis_list
        #     }
        # )

    def _print(self, *values):
        """Helper method for printing if ``verbose`` is set to True."""
        if self._print_verbose:
            print(*values)

    def _query_hf_data(self, trace: dict) -> Optional[dict]:
        """Query the internal ``_hf_data`` attribute and returns a match based on
        ``uid``.

        Parameters
        ----------
        trace : dict
            The trace where we want to find a match for.

        Returns
        -------
        Optional[dict]
            The ``hf_data``-trace dict if a match is found, else ``None``.

        """
        uid = trace["uid"]
        hf_trace_data = self._hf_data.get(uid)
        if hf_trace_data is None:
            trace_props = {
                k: trace[k] for k in set(trace.keys()).difference({"x", "y"})
            }
            self._print(f"[W] trace with {trace_props} not found")
        return hf_trace_data

    def _get_current_graph(self) -> dict:
        """Create an efficient copy of the current graph by omitting the "hovertext",
        "x", and "y" properties of each trace.

        Returns
        -------
        dict
            The current graph dict

        See Also
        --------
        https://github.com/plotly/plotly.py/blob/2e7f322c5ea4096ce6efe3b4b9a34d9647a8be9c/packages/python/plotly/plotly/basedatatypes.py#L3278
        """
        return {
            "data": [
                {
                    k: copy(trace[k])
                    # TODO: why not "text" as well? -> we can use _hf_data_container.fields then
                    for k in set(trace.keys()).difference({"x", "y", "hovertext"})
                }
                for trace in self._data
            ],
            "layout": copy(self._layout),
        }

    def _parse_trace_name(
        self, hf_trace_data: dict, slice_len: int, agg_x: np.ndarray
    ) -> str:
        """Parse the trace name.

        Parameters
        ----------
        hf_trace_data : dict
            The high-frequency trace data dict.
        slice_len : int
            The length of the slice.
        agg_x : np.ndarray
            The x-axis values of the aggregated trace.

        Returns
        -------
        str
            The parsed trace name.
            When no downsampling is needed, the original trace name is returned.
            When downsampling is needed, the average bin size (expressed in x-units) is
            added in orange color with a `~` to the trace name.

        """
        if slice_len <= hf_trace_data["max_n_samples"]:  # When no downsampling needed
            return hf_trace_data["name"]

        # The data is downsampled, so we add the downsampling information to the name
        agg_prefix, agg_suffix = ' <i style="color:#fc9944">~', "</i>"
        name = self._prefix + hf_trace_data["name"] + self._suffix

        # Add the mean aggregation bin size to the trace name
        if self._show_mean_aggregation_size:
            # Base case ...
            if len(agg_x) < 2:
                return name

            mean_bin_size = (agg_x[-1] - agg_x[0]) / agg_x.shape[0]  # mean bin size
            if isinstance(mean_bin_size, (np.timedelta64, pd.Timedelta)):
                mean_bin_size = round_td_str(pd.Timedelta(mean_bin_size))
            else:
                mean_bin_size = round_number_str(mean_bin_size)
            name += f"{agg_prefix}{mean_bin_size}{agg_suffix}"
        return name

    def _check_update_trace_data(
        self,
        trace: dict,
        start: Optional[Union[str, float]] = None,
        end: Optional[Union[str, float]] = None,
    ) -> Optional[Union[dict, BaseTraceType]]:
        """Check and update the passed ``trace`` its data properties based on the
        slice range.

        Note
        ----
        This is a pass by reference. The passed trace object will be updated and
        returned if found in ``hf_data``.

        Parameters
        ----------
        trace : BaseTraceType or dict
             - An instances of a trace class from the ``plotly.graph_objects`` (go)
                package (e.g, ``go.Scatter``, ``go.Bar``)
             - or a dict where:

                  - The 'type' property specifies the trace type (e.g.
                    'scatter', 'bar', 'area', etc.). If the dict has no 'type'
                    property then 'scatter' is assumed.
                  - All remaining properties are passed to the constructor
                    of the specified trace type.

        start : Union[float, str], optional
            The start index for which we want resampled data to be updated to,
            by default None,
        end : Union[float, str], optional
            The end index for which we want the resampled data to be updated to,
            by default None

        Returns
        -------
        Optional[Union[dict, BaseTraceType]]
            If the matching ``hf_series`` is found in ``hf_dict``, an (updated) trace
            will be returned, otherwise None.

        Note
        ----
        * If ``start`` and ``stop`` are strings, they most likely represent time-strings
        * ``start`` and ``stop`` will always be of the same type (float / time-string)
           because their underlying axis is the same.

        """
        hf_trace_data = self._query_hf_data(trace)

        if hf_trace_data is None:
            self._print("hf_data not found")
            return None

        # Parse trace data (necessary when updating the trace data)
        for k in _hf_data_container._fields:
            if isinstance(
                hf_trace_data[k], (np.ndarray, pd.RangeIndex, pd.DatetimeIndex)
            ):
                # is faster to escape the loop here than check inside the hasattr if
                continue
            elif pd.core.dtypes.common.is_datetime64tz_dtype(hf_trace_data[k]):
                # When we use the .values method, timezone information is lost
                # so convert it to pd.DatetimeIndex, which preserves the tz-info
                hf_trace_data[k] = pd.Index(hf_trace_data[k])
            elif hasattr(hf_trace_data[k], "values"):
                # when not a range index or datetime index
                hf_trace_data[k] = hf_trace_data[k].values

        # Also check if the y-data is empty, if so, return an empty trace
        if len(hf_trace_data["y"]) == 0:
            trace["x"] = []
            trace["y"] = []
            trace["name"] = hf_trace_data["name"]
            return trace

        # Leverage the axis type to get the start and end indices
        # Note: the axis type specified in the figure layout takes precedence over the
        # the axis type which is inferred from the data (and stored in hf_trace_data)
        # TODO: verify if we need to use `axis`of anchor as key to determing axis type
        axis = trace.get("xaxis", "x")
        axis_type = self.layout._props.get(axis[:1] + "axis" + axis[1:], {}).get(
            "type", hf_trace_data["axis_type"]
        )
        start_idx, end_idx = PlotlyAggregatorParser.get_start_end_indices(
            hf_trace_data, axis_type, start, end
        )

        # Return an invisible, single-point, trace when the sliced hf_series doesn't
        # contain any data in the current view
        if end_idx == start_idx:
            trace["x"] = [hf_trace_data["x"][0]]
            trace["y"] = [None]
            trace["name"] = hf_trace_data["name"]
            return trace

        agg_x, agg_y, indices = PlotlyAggregatorParser.aggregate(
            hf_trace_data, start_idx, end_idx
        )

        # -------------------- Set the hf_trace_data_props -------------------
        # Parse the data types to an orjson compatible format
        # NOTE: this can be removed once orjson supports f16
        trace["x"] = self._parse_dtype_orjson(agg_x)
        trace["y"] = self._parse_dtype_orjson(agg_y)
        trace["name"] = self._parse_trace_name(
            hf_trace_data, end_idx - start_idx, agg_x
        )

        def _nest_dict_rec(k: str, v: any, out: dict) -> None:
            """Recursively nest a dict based on the key whose '_' indicates level."""
            k, *rest = k.split("_", 1)
            if rest:
                _nest_dict_rec(rest[0], v, out.setdefault(k, {}))
            else:
                out[k] = v

        # Check if (hover)text also needs to be downsampled
        for k in ["text", "hovertext", "marker_size", "marker_color"]:
            k_val = hf_trace_data.get(k)
            if isinstance(k_val, (np.ndarray, pd.Series)):
                assert isinstance(
                    hf_trace_data["downsampler"], DataPointSelector
                ), "Only DataPointSelector can downsample non-data trace array props."
                _nest_dict_rec(k, k_val[start_idx + indices], trace)
            elif k_val is not None:
                trace[k] = k_val

        return trace

    def _check_update_figure_dict(
        self,
        figure: dict,
        start: Optional[Union[float, str]] = None,
        stop: Optional[Union[float, str]] = None,
        xaxis_filter: str = None,
        updated_trace_indices: Optional[List[int]] = None,
    ) -> List[int]:
        """Check and update the traces within the figure dict.

        hint
        ----
        This method will most likely be used within a ``Dash`` callback to resample the
        view, based on the configured number of parameters.

        Note
        ----
        This is a pass by reference. The passed figure object will be updated.
        No new view of this figure will be created, hence no return!

        Parameters
        ----------
        figure : dict
            The figure dict which will be updated.
        start : Union[float, str], optional
            The start time for the new resampled data view, by default None.
        stop : Union[float, str], optional
            The end time for the new resampled data view, by default None.
        xaxis_filter: str, optional
            Additional trace-update subplot filter, by default None.
        updated_trace_indices: List[int], optional
            List of trace indices that already have been updated, by default None.

        Returns
        -------
        List[int]
            A list of indices withholding the trace-data-array-index from the of data
            modalities which are updated.

        """
        xaxis_filter_short = None
        if xaxis_filter is not None:
            xaxis_filter_short = "x" + xaxis_filter.lstrip("xaxis")

        if updated_trace_indices is None:
            updated_trace_indices = []

        for idx, trace in enumerate(figure["data"]):
            # We skip when the trace-idx already has been updated.
            if idx in updated_trace_indices:
                continue

            if xaxis_filter is not None:
                # the x-anchor of the trace is stored in the layout data
                if trace.get("yaxis") is None:
                    # TODO In versions up until v0.8.2 we made the assumption that yaxis
                    # = xaxis_filter_short. -> Why did we make this assumption?
                    y_axis = "y"  # + xaxis_filter[1:]
                else:
                    y_axis = "yaxis" + trace.get("yaxis")[1:]

                # Next to the x-anchor, we also fetch the xaxis which matches the
                # current trace (i.e. if this value is not None, the axis shares the
                # x-axis with one or more traces).
                # This is relevant when e.g. fig.update_traces(xaxis='x...') was called.
                x_anchor_trace = figure["layout"].get(y_axis, {}).get("anchor")
                if x_anchor_trace is not None:
                    xaxis_matches = (
                        figure["layout"]
                        .get("xaxis" + x_anchor_trace.lstrip("x"), {})
                        .get("matches")
                    )
                else:
                    xaxis_matches = figure["layout"].get("xaxis", {}).get("matches")

                # print(
                #     f"x_anchor: {x_anchor_trace} - xaxis_filter: {xaxis_filter} ",
                #     f"- xaxis_matches: {xaxis_matches}"
                # )

                # We skip when:
                # * the change was made on the first row and the trace its anchor is not
                #   in [None, 'x'] and the matching (a.k.a. shared) xaxis is not equal
                #   to the xaxis filter argument.
                #   -> why None: traces without row/col argument and stand on first row
                #      and do not have the anchor property (hence the DICT.get() method)
                # * x_axis_filter_short not in [x_anchor or xaxis matches] for
                #   NON first rows
                if (
                    xaxis_filter_short == "x"
                    and (
                        x_anchor_trace not in (None, "x")
                        and xaxis_matches != xaxis_filter_short
                    )
                ) or (
                    xaxis_filter_short != "x"
                    and (xaxis_filter_short not in (x_anchor_trace, xaxis_matches))
                ):
                    continue

            # If we managed to find and update the trace, it will return the trace
            # and thus not None.
            updated_trace = self._check_update_trace_data(trace, start=start, end=stop)
            if updated_trace is not None:
                updated_trace_indices.append(idx)
        return updated_trace_indices

    @staticmethod
    def _get_figure_class(constr: type) -> type:
        """Get the plotly figure class (constructor) for the given class (constructor).

        !!! note
            This method will always return a plotly constructor, even when the given
            `constr` is decorated (after executing the ``register_plotly_resampler``
            function).

        Parameters
        ----------
        constr: type
            The constructor class for which we want to retrieve the plotly constructor.

        Returns
        -------
        type:
            The plotly figure class (constructor) of the given `constr`.

        """
        from ..registering import _get_plotly_constr  # To avoid ImportError

        return _get_plotly_constr(constr)

    @property
    def hf_data(self):
        """Property to adjust the `data` component of the current graph

        !!! note
            The user has full responsibility to adjust ``hf_data`` properly.


        ??? example

            ```python
                >>> from plotly_resampler import FigureResampler
                >>> fig = FigureResampler(go.Figure())
                >>> fig.add_trace(...)
                >>> # Adjust the y property of the above added trace
                >>> fig.hf_data[-1]["y"] = - s ** 2
                >>> fig.hf_data
                [
                    {
                        'max_n_samples': 1000,
                        'x': RangeIndex(start=0, stop=11000000, step=1),
                        'y': array([-0.01339909,  0.01390696,, ...,  0.25051913, 0.55876513]),
                        'axis_type': 'linear',
                        'downsampler': <plotly_resampler.aggregation.aggregators.LTTB at 0x7f786d5a9ca0>,
                        'text': None,
                        'hovertext': None
                    },
                ]
            ```
        """
        return list(self._hf_data.values())

    def _parse_get_trace_props(
        self,
        trace: BaseTraceType,
        hf_x: Iterable = None,
        hf_y: Iterable = None,
        hf_text: Iterable = None,
        hf_hovertext: Iterable = None,
        hf_marker_size: Iterable = None,
        hf_marker_color: Iterable = None,
        check_nans: bool = True,
    ) -> _hf_data_container:
        """Parse and capture the possibly high-frequency trace-props in a datacontainer.

        Parameters
        ----------
        trace : BaseTraceType
            The trace which will be parsed.
        hf_x : Iterable, optional
            High-frequency trace "x" data, overrides the current trace its x-data.
        hf_y : Iterable, optional
            High-frequency trace "y" data, overrides the current trace its y-data.
        hf_text : Iterable, optional
            High-frequency trace "text" data, overrides the current trace its text-data.
        hf_hovertext : Iterable, optional
            High-frequency trace "hovertext" data, overrides the current trace its
            hovertext data.
        check_nans: bool, optional
            Whether the `hf_y` should be checked for NaNs, by default True.
            As checking for NaNs is expensive, this can be disabled when the `hf_y` is
            already known to contain no NaNs (or when the downsampler can handle NaNs,
            e.g., EveryNthPoint).

        Returns
        -------
        _hf_data_container
            A namedtuple which serves as a datacontainer.

        """
        hf_x: np.ndarray = (
            # fmt: off
            (np.asarray(trace["x"]) if trace["x"] is not None else None)
            if hasattr(trace, "x") and hf_x is None
            # If we cast a tz-aware datetime64 array to `.values` we lose the tz-info 
            # and the UTC time will be displayed instead of the tz-localized time, 
            # hence we cast to a pd.DatetimeIndex, which preserves the tz-info
            # As a matter of fact, to resolve #231, we also convert non-tz-aware 
            # datetime64 arrays to an pd.Index
            else pd.Index(hf_x) if pd.core.dtypes.common.is_datetime64_any_dtype(hf_x)
            else hf_x.values if isinstance(hf_x, pd.Series)
            else hf_x if isinstance(hf_x, pd.Index)
            else np.asarray(hf_x)
            # fmt: on
        )

        hf_y = (
            trace["y"]
            if hasattr(trace, "y") and hf_y is None
            else hf_y.values
            if isinstance(hf_y, (pd.Series, pd.Index))
            else hf_y
        )
        # NOTE: the if will not be triggered for a categorical series its values
        if not hasattr(hf_y, "dtype"):
            hf_y: np.ndarray = np.asarray(hf_y)

        hf_text = (
            hf_text
            if hf_text is not None
            else trace["text"]
            if hasattr(trace, "text") and trace["text"] is not None
            else None
        )

        hf_hovertext = (
            hf_hovertext
            if hf_hovertext is not None
            else trace["hovertext"]
            if hasattr(trace, "hovertext") and trace["hovertext"] is not None
            else None
        )

        hf_marker_size = (
            trace["marker"]["size"]
            if (
                hf_marker_size is None
                and hasattr(trace, "marker")
                and "size" in trace["marker"]
            )
            else hf_marker_size
        )

        hf_marker_color = (
            trace["marker"]["color"]
            if (
                hf_marker_color is None
                and hasattr(trace, "marker")
                and "color" in trace["marker"]
            )
            else hf_marker_color
        )

        if trace["type"].lower() in self._high_frequency_traces:
            if hf_x is None:  # if no data as x or hf_x is passed
                if hf_y.ndim != 0:  # if hf_y is an array
                    hf_x = pd.RangeIndex(0, len(hf_y))  # np.arange(len(hf_y))
                else:  # if no data as y or hf_y is passed
                    hf_x = np.asarray(None)

            assert hf_y.ndim == np.ndim(hf_x), (
                "plotly-resampler requires scatter data "
                "(i.e., x and y, or hf_x and hf_y) to have the same dimensionality!"
            )
            # When the x or y of a trace has more than 1 dimension, it is not at all
            # straightforward how it should be resampled.
            assert hf_y.ndim <= 1 and np.ndim(hf_x) <= 1, (
                "plotly-resampler requires scatter data "
                "(i.e., x and y, or hf_x and hf_y) to be <= 1 dimensional!"
            )

            # Note: this converts the hf property to a np.ndarray
            if isinstance(hf_text, (tuple, list, np.ndarray, pd.Series)):
                hf_text = np.asarray(hf_text)
            if isinstance(hf_hovertext, (tuple, list, np.ndarray, pd.Series)):
                hf_hovertext = np.asarray(hf_hovertext)
            if isinstance(hf_marker_size, (tuple, list, np.ndarray, pd.Series)):
                hf_marker_size = np.asarray(hf_marker_size)
            if isinstance(hf_marker_color, (tuple, list, np.ndarray, pd.Series)):
                hf_marker_color = np.asarray(hf_marker_color)

            # Remove NaNs for efficiency (storing less meaningless data)
            # NaNs introduce gaps between enclosing non-NaN data points & might distort
            # the resampling algorithms
            if check_nans and pd.isna(hf_y).any():
                not_nan_mask = ~pd.isna(hf_y)
                hf_x = hf_x[not_nan_mask]
                hf_y = hf_y[not_nan_mask]
                if isinstance(hf_text, np.ndarray):
                    hf_text = hf_text[not_nan_mask]
                if isinstance(hf_hovertext, np.ndarray):
                    hf_hovertext = hf_hovertext[not_nan_mask]
                if isinstance(hf_marker_size, np.ndarray):
                    hf_marker_size = hf_marker_size[not_nan_mask]
                if isinstance(hf_marker_color, np.ndarray):
                    hf_marker_color = hf_marker_color[not_nan_mask]

            # Try to parse the hf_x data if it is of object type or
            if len(hf_x) and (hf_x.dtype.type is np.str_ or hf_x.dtype == "object"):
                try:
                    # Try to parse to numeric
                    hf_x = pd.to_numeric(hf_x, errors="raise")
                except (ValueError, TypeError):
                    try:
                        # Try to parse to datetime
                        hf_x = pd.to_datetime(hf_x, utc=False, errors="raise")
                        # Will be cast to object array if it contains multiple timezones.
                        if hf_x.dtype == "object":
                            raise ValueError(
                                "The x-data contains multiple timezones, which is not "
                                "supported by plotly-resampler!"
                            )
                    except (ValueError, TypeError):
                        raise ValueError(
                            "plotly-resampler requires the x-data to be numeric or "
                            "datetime-like \nMore details in the stacktrace above."
                        )

            # If the categorical or string-like hf_y data is of type object (happens
            # when y argument is used for the trace constructor instead of hf_y), we
            # transform it to type string as such it will be sent as categorical data
            # to the downsampling algorithm
            if hf_y.dtype == "object" or hf_y.dtype.type == np.str_:
                # But first, we try to parse to a numeric dtype (as this is the
                # behavior that plotly supports)
                # Note that a bool array of type object will remain a bool array (and
                # not will be transformed to an array of ints (0, 1))
                try:
                    hf_y = pd.to_numeric(hf_y, errors="raise")
                except ValueError:
                    hf_y = pd.Categorical(hf_y)  # TODO: ordered=True?
            assert len(hf_x) == len(hf_y), "x and y have different length!"
        else:
            self._print(f"trace {trace['type']} is not a high-frequency trace")

            # hf_x and hf_y have priority over the traces' data
            if hasattr(trace, "x"):
                trace["x"] = hf_x

            if hasattr(trace, "y"):
                trace["y"] = hf_y

            if hasattr(trace, "text"):
                trace["text"] = hf_text

            if hasattr(trace, "hovertext"):
                trace["hovertext"] = hf_hovertext
            if hasattr(trace, "marker"):
                if hasattr(trace.marker, "size"):
                    trace.marker.size = hf_marker_size
                if hasattr(trace.marker, "color"):
                    trace.marker.color = hf_marker_color

        return _hf_data_container(
            hf_x, hf_y, hf_text, hf_hovertext, hf_marker_size, hf_marker_color
        )

    def _construct_hf_data_dict(
        self,
        dc: _hf_data_container,
        trace: BaseTraceType,
        downsampler: AbstractAggregator | None,
        gap_handler: AbstractGapHandler | None,
        max_n_samples: int | None,
        offset=0,
    ) -> dict:
        """Create the `hf_data` dict which will be put in the `_hf_data` property.

        Parameters
        ----------
        dc : _hf_data_container
            The hf_data container, withholding the parsed hf-data.
        trace : BaseTraceType
            The trace.
        downsampler : AbstractAggregator | None
            The downsampler which will be used.
        gap_handler : AbstractGapHandler | None
            The gap handler which will be used.
        max_n_samples : int | None
            The max number of output samples.

        Returns
        -------
        dict
            The hf_data dict.
        """
        # Checking this now avoids less interpretable `KeyError` when resampling
        assert_text = (
            "In order to perform time series aggregation, the data must be "
            "sorted in time; i.e., the x-data must be (non-strictly) "
            "monotonically increasing."
        )
        if isinstance(dc.x, pd.Index):
            assert dc.x.is_monotonic_increasing, assert_text
        else:
            assert pd.Series(dc.x).is_monotonic_increasing, assert_text

        # As we support prefix-suffixing of downsampled data, we assure that
        # each trace has a name
        # https://github.com/plotly/plotly.py/blob/ce0ed07d872c487698bde9d52e1f1aadf17aa65f/packages/python/plotly/plotly/basedatatypes.py#L539
        # The link above indicates that the trace index is derived from `data`
        if trace.name is None:
            trace.name = f"trace {len(self.data) + offset}"

        # Determine (1) the axis type and (2) the downsampler instance
        # & (3) store a hf_data entry for the corresponding trace,
        # identified by its UUID
        axis_type = (
            "date"
            if isinstance(dc.x, pd.DatetimeIndex)
            or pd.core.dtypes.common.is_datetime64_any_dtype(dc.x)
            else "linear"
        )

        default_n_samples = False
        if max_n_samples is None:
            default_n_samples = True
            max_n_samples = self._global_n_shown_samples

        default_downsampler = False
        if downsampler is None:
            default_downsampler = True
            downsampler = self._global_downsampler

        default_gap_handler = False
        if gap_handler is None:
            default_gap_handler = True
            gap_handler = self._global_gap_handler

        # TODO -> can't we just store the DC here (might be less duplication of
        #  code knowledge, because now, you need to know all the eligible hf_keys in
        #  dc
        return {
            "max_n_samples": max_n_samples,
            "default_n_samples": default_n_samples,
            "name": trace.name,
            "axis_type": axis_type,
            "downsampler": downsampler,
            "default_downsampler": default_downsampler,
            "gap_handler": gap_handler,
            "default_gap_handler": default_gap_handler,
            **dc._asdict(),
        }

    @staticmethod
    def _add_trace_to_add_traces_kwargs(kwargs: dict) -> dict:
        """Convert the `add_trace` kwargs to the `add_traces` kwargs."""
        # The keywords that need to be converted to a list
        convert_keywords = ["row", "col", "secondary_y"]

        updated_kwargs = {}  # The updated kwargs (from `add_trace` to `add_traces`)
        for keyword in convert_keywords:
            value = kwargs.pop(keyword, None)
            if value is not None:
                updated_kwargs[f"{keyword}s"] = [value]
            else:
                updated_kwargs[f"{keyword}s"] = None

        return {**kwargs, **updated_kwargs}

    def add_trace(
        self,
        trace: Union[BaseTraceType, dict],
        max_n_samples: int = None,
        downsampler: AbstractAggregator = None,
        gap_handler: AbstractGapHandler = None,
        limit_to_view: bool = False,
        # Use these if you want some speedups (and are working with really large data)
        hf_x: Iterable = None,
        hf_y: Iterable = None,
        hf_text: Union[str, Iterable] = None,
        hf_hovertext: Union[str, Iterable] = None,
        hf_marker_size: Union[str, Iterable] = None,
        hf_marker_color: Union[str, Iterable] = None,
        check_nans: bool = True,
        **trace_kwargs,
    ):
        """Add a trace to the figure.

        Parameters
        ----------
        trace : BaseTraceType or dict
            Either:

              - An instances of a trace class from the ``plotly.graph_objects`` (go)
                package (e.g., ``go.Scatter``, ``go.Bar``)
              - or a dict where:

                - The type property specifies the trace type (e.g. scatter, bar,
                  area, etc.). If the dict has no 'type' property then scatter is
                  assumed.
                - All remaining properties are passed to the constructor
                  of the specified trace type.
        max_n_samples : int, optional
            The maximum number of samples that will be shown by the trace.\n
            !!! note
                If this variable is not set; ``_global_n_shown_samples`` will be used.
        downsampler: AbstractAggregator, optional
            The abstract series downsampler method.\n
            !!! note
                If this variable is not set, ``_global_downsampler`` will be used.
        gap_handler: AbstractGapHandler, optional
            The abstract series gap handler method.\n
            !!! note
                If this variable is not set, ``_global_gap_handler`` will be used.
        limit_to_view: boolean, optional
            If set to True, the trace's datapoints will be cut to the corresponding
            front-end view, even if the total number of samples is lower than
            ``max_n_samples``, By default False.\n
            Remark that setting this parameter to True ensures that low frequency traces
            are added to the ``hf_data`` property.
        hf_x: Iterable, optional
            The original high frequency series positions, can be either a time-series or
            an increasing, numerical index. If set, this has priority over the trace its
            data.
        hf_y: Iterable, optional
            The original high frequency values. If set, this has priority over the
            trace its data.
        hf_text: Iterable, optional
            The original high frequency text. If set, this has priority over the trace
            its ``text`` argument.
        hf_hovertext: Iterable, optional
            The original high frequency hovertext. If set, this has priority over the
            trace its ```hovertext`` argument.
        hf_marker_size: Iterable, optional
            The original high frequency marker size. If set, this has priority over the
            trace its ``marker.size`` argument.
        hf_marker_color: Iterable, optional
            The original high frequency marker color. If set, this has priority over the
            trace its ``marker.color`` argument.
        check_nans: boolean, optional
            If set to True, the trace's data will be checked for NaNs - which will be
            removed. By default True.
            As this is a costly operation, it is recommended to set this parameter to
            False if you are sure that your data does not contain NaNs (or when the
            downsampler can handle NaNs, e.g., EveryNthPoint). This should considerably
            speed up the graph construction time.
        **trace_kwargs: dict
            Additional trace related keyword arguments.
            e.g.: row=.., col=..., secondary_y=...

            !!! info "See Also"
                [`Figure.add_trace`](https://plotly.com/python-api-reference/generated/plotly.graph_objects.Figure.html#plotly.graph_objects.Figure.add_trace>) docs.

        Returns
        -------
        BaseFigure
            The Figure on which ``add_trace`` was called on; i.e. self.

        !!! note

            Constructing traces with **very large data amounts** really takes some time.
            To speed this up; use this [`add_trace`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_trace] method and

            1. Create a trace with no data (empty lists)
            2. pass the high frequency data to this method using the ``hf_x`` and ``hf_y``
               parameters.

            See the example below:
                ```python
                >>> from plotly.subplots import make_subplots
                >>> s = pd.Series()  # a high-frequency series, with more than 1e7 samples
                >>> fig = FigureResampler(go.Figure())
                >>> fig.add_trace(go.Scattergl(x=[], y=[], ...), hf_x=s.index, hf_y=s)
                ```

            !!! todo
                * explain why adding x and y to a trace is so slow
                * check and simplify the example above

        !!! tip

            * If you **do not want to downsample** your data, set ``max_n_samples`` to the
              the number of datapoints of your trace!

        !!! warning

            * The ``NaN`` values in either ``hf_y`` or ``trace.y`` will be omitted! We do
              not allow ``NaN`` values in ``hf_x`` or ``trace.x``.
            * ``hf_x``, ``hf_y``, ``hf_text``, and ``hf_hovertext`` are useful when you deal
              with large amounts of data (as it can increase the speed of this add_trace()
              method with ~30%). These arguments have priority over the trace's data and
              (hover)text attributes.
            * Low-frequency time-series data, i.e. traces that are not resampled, can hinder
              the the automatic-zooming (y-scaling) as these will not be stored in the
              back-end and thus not be scaled to the view.
              To circumvent this, the ``limit_to_view`` argument can be set, resulting in
              also storing the low-frequency series in the back-end.

        """
        # to comply with the plotly data input acceptance behavior
        if isinstance(trace, (list, tuple)):
            raise ValueError("Trace must be either a dict or a BaseTraceType")

        max_out_s = (
            self._global_n_shown_samples if max_n_samples is None else max_n_samples
        )

        # Validate the trace and convert to a trace object
        if not isinstance(trace, BaseTraceType):
            trace = self._data_validator.validate_coerce(trace)[0]

        # First add a UUID, as each (even the non-hf_data traces), must contain this
        # key for comparison. If the trace already has a UUID, we will keep it.
        uuid_str = str(uuid4()) if trace.uid is None else trace.uid
        trace.uid = uuid_str

        # construct the hf_data_container
        # TODO in future version -> maybe regex on kwargs which start with `hf_`
        dc = self._parse_get_trace_props(
            trace,
            hf_x,
            hf_y,
            hf_text,
            hf_hovertext,
            hf_marker_size,
            hf_marker_color,
            check_nans,
        )

        # These traces will determine the autoscale its RANGE!
        #   -> so also store when `limit_to_view` is set.
        if trace["type"].lower() in self._high_frequency_traces:
            n_samples = len(dc.x)
            if n_samples > max_out_s or limit_to_view:
                self._print(
                    f"\t[i] DOWNSAMPLE {trace['name']}\t{n_samples}->{max_out_s}"
                )

                self._hf_data[uuid_str] = self._construct_hf_data_dict(
                    dc,
                    trace=trace,
                    downsampler=downsampler,
                    gap_handler=gap_handler,
                    max_n_samples=max_n_samples,
                )

                # Before we update the trace, we create a new pointer to that trace in
                # which the downsampled data will be stored. This way, the original
                # data of the trace to this `add_trace` method will not be altered.
                # We copy (by reference) all the non-data properties of the trace in
                # the new trace.
                trace = trace._props  # convert the trace into a dict
                # NOTE: there is no need to store `marker` property here.
                # If needed, it will be added  to `trace` via `check_update_trace_data`
                trace = {
                    k: trace[k] for k in set(trace.keys()).difference(set(dc._fields))
                }

                # NOTE:
                # If all the raw data needs to be sent to the javascript, and the trace
                # is high-frequency, this would take significant time!
                # Hence, you first downsample the trace.
                trace = self._check_update_trace_data(trace)
                assert trace is not None
                return super(AbstractFigureAggregator, self).add_traces(
                    [trace], **self._add_trace_to_add_traces_kwargs(trace_kwargs)
                )
            else:
                self._print(f"[i] NOT resampling {trace['name']} - len={n_samples}")
                trace._process_kwargs(**{k: getattr(dc, k) for k in dc._fields})
                return super(AbstractFigureAggregator, self).add_traces(
                    [trace], **self._add_trace_to_add_traces_kwargs(trace_kwargs)
                )

        return super(self._figure_class, self).add_traces(
            [trace], **self._add_trace_to_add_traces_kwargs(trace_kwargs)
        )

    def add_traces(
        self,
        data: List[BaseTraceType | dict] | BaseTraceType | Dict,
        max_n_samples: None | List[int] | int = None,
        downsamplers: None | List[AbstractAggregator] | AbstractAggregator = None,
        gap_handlers: None | List[AbstractGapHandler] | AbstractGapHandler = None,
        limit_to_views: List[bool] | bool = False,
        check_nans: List[bool] | bool = True,
        **traces_kwargs,
    ):
        """Add traces to the figure.

        !!! note

            Make sure to look at the [`add_trace`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_trace] function for more info about
            **speed optimization**, and dealing with not ``high-frequency`` data, but
            still want to resample / limit the data to the front-end view.

        Parameters
        ----------
        data : List[BaseTraceType  |  dict]
            A list of trace specifications to be added.
            Trace specifications may be either:

              - Instances of trace classes from the plotly.graph_objs
                package (e.g plotly.graph_objs.Scatter, plotly.graph_objs.Bar).
              - Dicts where:

                  - The 'type' property specifies the trace type (e.g.
                    'scatter', 'bar', 'area', etc.). If the dict has no 'type'
                    property then 'scatter' is assumed.
                  - All remaining properties are passed to the constructor
                    of the specified trace type.
        max_n_samples : None | List[int] | int, optional
              The maximum number of samples that will be shown for each trace.
              If a single integer is passed, all traces will use this number. If this
              variable is not set; ``_global_n_shown_samples`` will be used.
        downsamplers : None | List[AbstractAggregator] | AbstractAggregator, optional
            The downsampler that will be used to aggregate the traces. If a single
            aggregator is passed, all traces will use this aggregator.
            If this variable is not set, ``_global_downsampler`` will be used.
        gap_handlers : None | List[AbstractGapHandler] | AbstractGapHandler, optional
            The gap handler that will be used to aggregate the traces. If a single
            gap handler is passed, all traces will use this gap handler.
            If this variable is not set, ``_global_gap_handler`` will be used.
        limit_to_views : None | List[bool] | bool, optional
            List of limit_to_view booleans for the added traces. If set to True the
            trace's datapoints will be cut to the corresponding front-end view, even if
            the total number of samples is lower than ``max_n_samples``.
            If a single boolean is passed, all to be added traces will use this value,
            by default False.\n
            Remark that setting this parameter to True ensures that low frequency traces
            are added to the ``hf_data`` property.
        check_nans : None | List[bool] | bool, optional
            List of check_nans booleans for the added traces. If set to True, the
            trace's datapoints will be checked for NaNs. If a single boolean is passed,
            all to be added traces will use this value, by default True.\n
            As this is a costly operation, it is recommended to set this parameter to
            False if the data is known to contain no NaNs (or when the downsampler can
            handle NaNs, e.g., EveryNthPoint). This will considerably speed up the graph
            construction time.
        **traces_kwargs: dict
            Additional trace related keyword arguments.
            e.g.: rows=.., cols=..., secondary_ys=...

            !!! info "See Also"

                [`Figure.add_traces`](https://plotly.com/python-api-reference/generated/plotly.graph_objects.Figure.html#plotly.graph_objects.Figure.add_traces>) docs.

        Returns
        -------
        BaseFigure
            The Figure on which ``add_traces`` was called on; i.e. self.

        """
        # note: Plotly its add_traces also a allows non list-like input e.g. a scatter
        # object; the code below is an exact copy of their internally applied parsing
        if not isinstance(data, (list, tuple)):
            data = [data]

        # Convert each trace into a BaseTraceType object
        data = [
            self._data_validator.validate_coerce(trace)[0]
            if not isinstance(trace, BaseTraceType)
            else trace
            for trace in data
        ]

        # First add a UUID, as each (even the non-hf_data traces), must contain this
        # key for comparison. If the trace already has a UUID, we will keep it.
        for trace in data:
            uuid_str = str(uuid4()) if trace.uid is None else trace.uid
            trace.uid = uuid_str

        # Convert the data properties
        if isinstance(max_n_samples, (int, np.integer)) or max_n_samples is None:
            max_n_samples = [max_n_samples] * len(data)
        if isinstance(downsamplers, AbstractAggregator) or downsamplers is None:
            downsamplers = [downsamplers] * len(data)
        if isinstance(gap_handlers, AbstractGapHandler) or gap_handlers is None:
            gap_handlers = [gap_handlers] * len(data)
        if isinstance(limit_to_views, bool):
            limit_to_views = [limit_to_views] * len(data)
        if isinstance(check_nans, bool):
            check_nans = [check_nans] * len(data)

        zipped = zip(
            data, max_n_samples, downsamplers, gap_handlers, limit_to_views, check_nans
        )
        for (
            i,
            (trace, max_out, downsampler, gap_handler, limit_to_view, check_nan),
        ) in enumerate(zipped):
            if (
                trace.type.lower() not in self._high_frequency_traces
                or self._hf_data.get(trace.uid) is not None
            ):
                continue

            max_out_s = self._global_n_shown_samples if max_out is None else max_out
            if not limit_to_view and (trace.y is None or len(trace.y) <= max_out_s):
                continue

            dc = self._parse_get_trace_props(trace, check_nans=check_nan)
            self._hf_data[trace.uid] = self._construct_hf_data_dict(
                dc,
                trace=trace,
                downsampler=downsampler,
                gap_handler=gap_handler,
                max_n_samples=max_out,
                offset=i,
            )

            # convert the trace into a dict, and only withholds the non-hf props
            trace = trace._props
            trace = {k: trace[k] for k in set(trace.keys()).difference(set(dc._fields))}

            # update the trace data with the HF props
            trace = self._check_update_trace_data(trace)
            assert trace is not None
            data[i] = trace

        return super(self._figure_class, self).add_traces(data, **traces_kwargs)

    def _clear_figure(self):
        """Clear the current figure object its data and layout."""
        self._hf_data = {}
        self.data = []
        self._data = []
        self._layout = {}
        self.layout = {}

    def _copy_hf_data(self, hf_data: dict, adjust_default_values: bool = False) -> dict:
        """Copy (i.e. create a new key reference, not a deep copy) of a hf_data dict.

        Parameters
        ----------
        hf_data : dict
            The hf_data dict, having the trace 'uid' as key and the
            hf-data, together with its aggregation properties as dict-values
        adjust_default_values: bool
            Whether the default values (of the downsampler, max # shown samples) will
            be adjusted according to the values of this object, by default False

        Returns
        -------
        dict
            The copied (& default values adjusted) output dict.

        """
        hf_data_cp = {
            uid: {k: hf_dict[k] for k in set(hf_dict.keys())}
            for uid, hf_dict in hf_data.items()
        }

        # Adjust the default arguments to the current argument values
        if adjust_default_values:
            for hf_props in hf_data_cp.values():
                if hf_props.get("default_downsampler", False):
                    hf_props["downsampler"] = self._global_downsampler
                if hf_props.get("default_gap_handler", False):
                    hf_props["gap_handler"] = self._global_gap_handler
                if hf_props.get("default_n_samples", False):
                    hf_props["max_n_samples"] = self._global_n_shown_samples

        return hf_data_cp

    def replace(self, figure: go.Figure, convert_existing_traces: bool = True):
        """Replace the current figure layout with the passed figure object.

        Parameters
        ----------
        figure: go.Figure
            The figure object which will replace the existing figure.
        convert_existing_traces: bool, Optional
            A bool indicating whether the traces of the passed ``figure`` should be
            resampled, by default True.

        """
        self._clear_figure()
        self.__init__(
            figure=figure,
            convert_existing_traces=convert_existing_traces,
            default_n_shown_samples=self._global_n_shown_samples,
            default_downsampler=self._global_downsampler,
            default_gap_handler=self._global_gap_handler,
            resampled_trace_prefix_suffix=(self._prefix, self._suffix),
            show_mean_aggregation_size=self._show_mean_aggregation_size,
            verbose=self._print_verbose,
        )

    def _parse_relayout(self, relayout_dict: dict) -> dict:
        """Update the relayout object so that the autorange will be set to None when
        there are xy-matches.

        Parameters
        ----------
        relayout_dict : dict
            The relayout dictionary.
        """
        # 1. Create a new dict with additional layout updates for the front-end
        extra_layout_updates = {}

        # 1.1. Set autorange to False for each layout item with a specified x-range
        xy_matches = self._re_matches(
            re.compile(r"[xy]axis\d*.range\[\d+]"), relayout_dict.keys()
        )
        for range_change_axis in xy_matches:
            axis = range_change_axis.split(".")[0]
            extra_layout_updates[f"{axis}.autorange"] = None
        return extra_layout_updates

    def construct_update_data(
        self,
        relayout_data: dict,
    ) -> Union[List[dict], dash.no_update]:
        """Construct the to-be-updated front-end data, based on the layout change.

        Attention
        ---------
        This method is tightly coupled with Dash app callbacks. It takes the front-end
        figure its ``relayoutData`` as input and returns the data which needs to be
        sent tot the ``TraceUpdater`` its ``updateData`` property for that corresponding
        graph.

        Parameters
        ----------
        relayout_data: dict
            A dict containing the ``relayout``-data (a.k.a. changed layout data) of
            the corresponding front-end graph.

        Returns
        -------
        List[dict]:
            A list of dicts, where each dict-item is a representation of a trace its
            *data* properties which are affected by the front-end layout change. |br|
            In other words, only traces which need to be updated will be sent to the
            front-end. Additionally, each trace-dict withholds the *index* of its
            corresponding position in the ``figure[data]`` array with the ``index``-key
            in each dict.

        """
        current_graph = self._get_current_graph()
        updated_trace_indices, cl_k = [], []
        if relayout_data:
            self._print("-" * 100 + "\n", "changed layout", relayout_data)

            cl_k = list(relayout_data.keys())

            # ------------------ HF DATA aggregation ---------------------
            # 1. Base case - there is an x-range specified in the front-end
            start_matches = self._re_matches(re.compile(r"xaxis\d*.range\[0]"), cl_k)
            stop_matches = self._re_matches(re.compile(r"xaxis\d*.range\[1]"), cl_k)
            if start_matches and stop_matches:  # when both are not empty
                for t_start_key, t_stop_key in zip(start_matches, stop_matches):
                    # Check if the xaxis<NUMB> part of xaxis<NUMB>.[0-1] matches
                    xaxis = t_start_key.split(".")[0]
                    assert xaxis == t_stop_key.split(".")[0]
                    # -> we want to copy the layout on the back-end
                    updated_trace_indices = self._check_update_figure_dict(
                        current_graph,
                        start=relayout_data[t_start_key],
                        stop=relayout_data[t_stop_key],
                        xaxis_filter=xaxis,
                        updated_trace_indices=updated_trace_indices,
                    )

            # 2. The user clicked on either autorange | reset axes
            autorange_matches = self._re_matches(
                re.compile(r"xaxis\d*.autorange"), cl_k
            )
            spike_matches = self._re_matches(re.compile(r"xaxis\d*.showspikes"), cl_k)
            # 2.1 Reset-axes -> autorange & reset to the global data view
            if autorange_matches and spike_matches:  # when both are not empty
                for autorange_key in autorange_matches:
                    if relayout_data[autorange_key]:
                        xaxis = autorange_key.split(".")[0]
                        updated_trace_indices = self._check_update_figure_dict(
                            current_graph,
                            xaxis_filter=xaxis,
                            updated_trace_indices=updated_trace_indices,
                        )
            # 2.1. Autorange -> do nothing, the autorange will be applied on the
            #      current front-end view
            elif (
                autorange_matches and not spike_matches
            ):  # when only autorange is not empty
                # PreventUpdate returns a 204 status code response on the
                # relayout post request
                return dash.no_update

        # If we do not have any traces to be updated, we will return an empty
        # request response
        if not updated_trace_indices:  # when updated_trace_indices is empty
            # PreventUpdate returns a 204 status-code response on the relayout post
            # request
            return dash.no_update

        # -------------------- construct callback data --------------------------
        # 1. Create the layout data for the front-end
        layout_traces_list: List[dict] = [relayout_data]

        # 2. Create the additional trace data for the frond-end
        relevant_keys = list(_hf_data_container._fields) + ["name", "marker"]
        # Note that only updated trace-data will be sent to the client
        for idx in updated_trace_indices:
            trace = current_graph["data"][idx]
            # TODO: check if we can reduce even more
            trace_reduced = {k: trace[k] for k in relevant_keys if k in trace}

            # Store the index into the corresponding to-be-sent trace-data so
            # the client front-end can know which trace needs to be updated
            trace_reduced.update({"index": idx})
            layout_traces_list.append(trace_reduced)
        return layout_traces_list

    @staticmethod
    def _parse_dtype_orjson(series: np.ndarray) -> np.ndarray:
        """Verify the orjson compatibility of the series and convert it if needed."""
        # NOTE:
        #    * float16 and float128 aren't supported with latest orjson versions (3.8.1)
        #    * this method assumes that the it will not get a float128 series
        # -> this method can be removed if orjson supports float16
        if series.dtype == np.float16:
            return series.astype(np.float32)
        return series

    @staticmethod
    def _re_matches(regex: re.Pattern, strings: Iterable[str]) -> List[str]:
        """Returns all the items in ``strings`` which regex.match(es) ``regex``."""
        matches = []
        for item in strings:
            m = regex.match(item)
            if m is not None:
                matches.append(m.string)
        return sorted(matches)

    @staticmethod
    def _is_no_update(update_data: Union[List[dict], dash.no_update]) -> bool:
        return update_data is dash.no_update

    # ------------------------------- Magic methods ---------------------------------

    def _get_pr_props_keys(self) -> List[str]:
        """Returns the keys (i.e., the names) of the plotly-resampler properties.

        Note
        ----
        This method is used to serialize the object in the `__reduce__` method.

        """
        return [
            "_hf_data",
            "_global_n_shown_samples",
            "_print_verbose",
            "_show_mean_aggregation_size",
            "_prefix",
            "_suffix",
            "_global_downsampler",
            "_global_gap_handler",
        ]

    def __reduce__(self):
        """Overwrite the reduce method (which is used to support deep copying and
        pickling).

        Note
        ----
        We do not overwrite the `to_dict` method, as this is used to send the figure
        to the frontend (and thus should not capture the plotly-resampler properties).
        """
        _, props = super().__reduce__()
        assert len(props) == 1  # I don't know why this would be > 1
        props = props[0]

        # Add the plotly-resampler properties
        props["pr_props"] = {}
        for k in self._get_pr_props_keys():
            props["pr_props"][k] = getattr(self, k)
        return self.__class__, (props,)  # (props,) to comply with plotly magic

hf_data property

Property to adjust the data component of the current graph

Note

The user has full responsibility to adjust hf_data properly.

Example

    >>> from plotly_resampler import FigureResampler
    >>> fig = FigureResampler(go.Figure())
    >>> fig.add_trace(...)
    >>> # Adjust the y property of the above added trace
    >>> fig.hf_data[-1]["y"] = - s ** 2
    >>> fig.hf_data
    [
        {
            'max_n_samples': 1000,
            'x': RangeIndex(start=0, stop=11000000, step=1),
            'y': array([-0.01339909,  0.01390696,, ...,  0.25051913, 0.55876513]),
            'axis_type': 'linear',
            'downsampler': <plotly_resampler.aggregation.aggregators.LTTB at 0x7f786d5a9ca0>,
            'text': None,
            'hovertext': None
        },
    ]

__init__(figure, convert_existing_traces=True, default_n_shown_samples=1000, default_downsampler=MinMaxLTTB(), default_gap_handler=MedDiffGapHandler(), resampled_trace_prefix_suffix=('<b style="color:sandybrown">[R]</b> ', ''), show_mean_aggregation_size=True, convert_traces_kwargs=None, verbose=False)

Instantiate a resampling data mirror.

Parameters:

Name	Type	Description	Default
figure	BaseFigure	The figure that will be decorated. Can be either an empty figure (e.g., go.Figure(), make_subplots(), go.FigureWidget) or an existing figure.	required
convert_existing_traces	bool	A bool indicating whether the high-frequency traces of the passed figure should be resampled, by default True. Hence, when set to False, the high-frequency traces of the passed figure will not be resampled.	True
default_n_shown_samples	int	The default number of samples that will be shown for each trace, by default 1000. Note This can be overridden within the add_trace method. If a trace withholds fewer datapoints than this parameter, the data will not be aggregated.	1000
default_downsampler	AbstractAggregator	An instance which implements the AbstractSeriesDownsampler interface and will be used as default downsampler, by default MinMaxLTTB. Note This can be overridden within the add_trace method.	MinMaxLTTB()
default_gap_handler	AbstractGapHandler	An instance which implements the AbstractGapHandler interface and will be used as default gap handler, by default MedDiffGapHandler. Note This can be overridden within the add_trace method.	MedDiffGapHandler()
resampled_trace_prefix_suffix	Tuple[str, str]	A tuple which contains the prefix and suffix, respectively, which will be added to the trace its legend-name when a resampled version of the trace is shown. By default, a bold, orange [R] is shown as prefix (no suffix is shown).	('<b style="color:sandybrown">[R]</b> ', '')
show_mean_aggregation_size	bool	Whether the mean aggregation bin size will be added as a suffix to the trace its legend-name, by default True.	True
convert_traces_kwargs	dict | None	A dict of kwargs that will be passed to the add_traces method and will be used to convert the existing traces. Note This argument is only used when the passed figure contains data and convert_existing_traces is set to True.	None
verbose	bool	Whether some verbose messages will be printed or not, by default False.	False

Source code in plotly_resampler/figure_resampler/figure_resampler_interface.py

def __init__(
    self,
    figure: BaseFigure,
    convert_existing_traces: bool = True,
    default_n_shown_samples: int = 1000,
    default_downsampler: AbstractAggregator = MinMaxLTTB(),
    default_gap_handler: AbstractGapHandler = MedDiffGapHandler(),
    resampled_trace_prefix_suffix: Tuple[str, str] = (
        '<b style="color:sandybrown">[R]</b> ',
        "",
    ),
    show_mean_aggregation_size: bool = True,
    convert_traces_kwargs: dict | None = None,
    verbose: bool = False,
):
    """Instantiate a resampling data mirror.

    Parameters
    ----------
    figure: BaseFigure
        The figure that will be decorated. Can be either an empty figure
        (e.g., ``go.Figure()``, ``make_subplots()``, ``go.FigureWidget``) or an
        existing figure.
    convert_existing_traces: bool
        A bool indicating whether the high-frequency traces of the passed ``figure``
        should be resampled, by default True. Hence, when set to False, the
        high-frequency traces of the passed ``figure`` will not be resampled.
    default_n_shown_samples: int, optional
        The default number of samples that will be shown for each trace,
        by default 1000.\n
        !!! note
            * This can be overridden within the [`add_trace`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_trace] method.
            * If a trace withholds fewer datapoints than this parameter,
              the data will *not* be aggregated.
    default_downsampler: AbstractAggregator
        An instance which implements the AbstractSeriesDownsampler interface and
        will be used as default downsampler, by default ``MinMaxLTTB``. \n
        !!! note
            This can be overridden within the [`add_trace`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_trace] method.
    default_gap_handler: GapHandler
        An instance which implements the AbstractGapHandler interface and will be
        used as default gap handler, by default ``MedDiffGapHandler``. \n
        !!! note
            This can be overridden within the [`add_trace`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_trace] method.
    resampled_trace_prefix_suffix: str, optional
        A tuple which contains the ``prefix`` and ``suffix``, respectively, which
        will be added to the trace its legend-name when a resampled version of the
        trace is shown. By default, a bold, orange ``[R]`` is shown as prefix
        (no suffix is shown).
    show_mean_aggregation_size: bool, optional
        Whether the mean aggregation bin size will be added as a suffix to the trace
        its legend-name, by default True.
    convert_traces_kwargs: dict, optional
        A dict of kwargs that will be passed to the [`add_traces`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_traces] method and
        will be used to convert the existing traces. \n
        !!! note
            This argument is only used when the passed ``figure`` contains data and
            ``convert_existing_traces`` is set to True.
    verbose: bool, optional
        Whether some verbose messages will be printed or not, by default False.

    """
    self._hf_data: Dict[str, dict] = {}
    self._global_n_shown_samples = default_n_shown_samples
    self._print_verbose = verbose
    self._show_mean_aggregation_size = show_mean_aggregation_size

    assert len(resampled_trace_prefix_suffix) == 2
    self._prefix, self._suffix = resampled_trace_prefix_suffix

    self._global_downsampler = default_downsampler
    self._global_gap_handler = default_gap_handler

    # Given figure should always be a BaseFigure that is not wrapped by
    # a plotly-resampler class
    assert isinstance(figure, BaseFigure)
    assert not issubclass(type(figure), AbstractFigureAggregator)
    self._figure_class = figure.__class__

    # Overwrite the passed arguments with the property dict values
    # (this is the case when the PR figure is created from a pickled object)
    if hasattr(figure, "_pr_props"):
        pr_props = figure._pr_props  # a dict of PR properties
        if pr_props is not None:
            # Overwrite the default arguments with the serialized properties
            for k, v in pr_props.items():
                setattr(self, k, v)
        delattr(figure, "_pr_props")  # should not be stored anymore

    if convert_existing_traces:
        # call __init__ with the correct layout and set the `_grid_ref` of the
        # to-be-converted figure
        f_ = self._figure_class(layout=figure.layout)
        f_._grid_str = figure._grid_str
        f_._grid_ref = figure._grid_ref
        super().__init__(f_)

        if convert_traces_kwargs is None:
            convert_traces_kwargs = {}

        # make sure that the UIDs of these traces do not get adjusted
        self._data_validator.set_uid = False
        self.add_traces(figure.data, **convert_traces_kwargs)
    else:
        super().__init__(figure)
        self._data_validator.set_uid = False

    # A list of al xaxis and yaxis string names
    # e.g., "xaxis", "xaxis2", "xaxis3", .... for _xaxis_list
    self._xaxis_list = self._re_matches(re.compile("xaxis\d*"), self._layout.keys())
    self._yaxis_list = self._re_matches(re.compile("yaxis\d*"), self._layout.keys())
    # edge case: an empty `go.Figure()` does not yet contain axes keys
    if not len(self._xaxis_list):
        self._xaxis_list = ["xaxis"]
        self._yaxis_list = ["yaxis"]

__reduce__()

Overwrite the reduce method (which is used to support deep copying and pickling).

Note

We do not overwrite the to_dict method, as this is used to send the figure to the frontend (and thus should not capture the plotly-resampler properties).

Source code in plotly_resampler/figure_resampler/figure_resampler_interface.py

def __reduce__(self):
    """Overwrite the reduce method (which is used to support deep copying and
    pickling).

    Note
    ----
    We do not overwrite the `to_dict` method, as this is used to send the figure
    to the frontend (and thus should not capture the plotly-resampler properties).
    """
    _, props = super().__reduce__()
    assert len(props) == 1  # I don't know why this would be > 1
    props = props[0]

    # Add the plotly-resampler properties
    props["pr_props"] = {}
    for k in self._get_pr_props_keys():
        props["pr_props"][k] = getattr(self, k)
    return self.__class__, (props,)  # (props,) to comply with plotly magic

add_trace(trace, max_n_samples=None, downsampler=None, gap_handler=None, limit_to_view=False, hf_x=None, hf_y=None, hf_text=None, hf_hovertext=None, hf_marker_size=None, hf_marker_color=None, check_nans=True, **trace_kwargs)

Add a trace to the figure.

Parameters:

Name	Type	Description	Default
trace	BaseTraceType or dict	Either: An instances of a trace class from the plotly.graph_objects (go) package (e.g., go.Scatter, go.Bar) or a dict where: The type property specifies the trace type (e.g. scatter, bar, area, etc.). If the dict has no ‘type’ property then scatter is assumed. All remaining properties are passed to the constructor of the specified trace type.	required
max_n_samples	int	The maximum number of samples that will be shown by the trace. Note If this variable is not set; _global_n_shown_samples will be used.	None
downsampler	AbstractAggregator	The abstract series downsampler method. Note If this variable is not set, _global_downsampler will be used.	None
gap_handler	AbstractGapHandler	The abstract series gap handler method. Note If this variable is not set, _global_gap_handler will be used.	None
limit_to_view	bool	If set to True, the trace’s datapoints will be cut to the corresponding front-end view, even if the total number of samples is lower than max_n_samples, By default False. Remark that setting this parameter to True ensures that low frequency traces are added to the hf_data property.	False
hf_x	Iterable	The original high frequency series positions, can be either a time-series or an increasing, numerical index. If set, this has priority over the trace its data.	None
hf_y	Iterable	The original high frequency values. If set, this has priority over the trace its data.	None
hf_text	Union[str, Iterable]	The original high frequency text. If set, this has priority over the trace its text argument.	None
hf_hovertext	Union[str, Iterable]	The original high frequency hovertext. If set, this has priority over the trace its `hovertext argument.	None
hf_marker_size	Union[str, Iterable]	The original high frequency marker size. If set, this has priority over the trace its marker.size argument.	None
hf_marker_color	Union[str, Iterable]	The original high frequency marker color. If set, this has priority over the trace its marker.color argument.	None
check_nans	bool	If set to True, the trace’s data will be checked for NaNs - which will be removed. By default True. As this is a costly operation, it is recommended to set this parameter to False if you are sure that your data does not contain NaNs (or when the downsampler can handle NaNs, e.g., EveryNthPoint). This should considerably speed up the graph construction time.	True
**trace_kwargs		Additional trace related keyword arguments. e.g.: row=.., col=…, secondary_y=… See Also Figure.add_trace docs.	{}

Returns:

Type	Description
BaseFigure	The Figure on which add_trace was called on; i.e. self.

Note

Constructing traces with very large data amounts really takes some time. To speed this up; use this add_trace method and

1. Create a trace with no data (empty lists)
2. pass the high frequency data to this method using the hf_x and hf_y parameters.

See the example below:

>>> from plotly.subplots import make_subplots
>>> s = pd.Series()  # a high-frequency series, with more than 1e7 samples
>>> fig = FigureResampler(go.Figure())
>>> fig.add_trace(go.Scattergl(x=[], y=[], ...), hf_x=s.index, hf_y=s)

Todo

• explain why adding x and y to a trace is so slow
• check and simplify the example above

Tip

• If you do not want to downsample your data, set max_n_samples to the the number of datapoints of your trace!

Warning

• The NaN values in either hf_y or trace.y will be omitted! We do not allow NaN values in hf_x or trace.x.
• hf_x, hf_y, hf_text, and hf_hovertext are useful when you deal with large amounts of data (as it can increase the speed of this add_trace() method with ~30%). These arguments have priority over the trace’s data and (hover)text attributes.
• Low-frequency time-series data, i.e. traces that are not resampled, can hinder the the automatic-zooming (y-scaling) as these will not be stored in the back-end and thus not be scaled to the view. To circumvent this, the limit_to_view argument can be set, resulting in also storing the low-frequency series in the back-end.

Source code in plotly_resampler/figure_resampler/figure_resampler_interface.py

def add_trace(
    self,
    trace: Union[BaseTraceType, dict],
    max_n_samples: int = None,
    downsampler: AbstractAggregator = None,
    gap_handler: AbstractGapHandler = None,
    limit_to_view: bool = False,
    # Use these if you want some speedups (and are working with really large data)
    hf_x: Iterable = None,
    hf_y: Iterable = None,
    hf_text: Union[str, Iterable] = None,
    hf_hovertext: Union[str, Iterable] = None,
    hf_marker_size: Union[str, Iterable] = None,
    hf_marker_color: Union[str, Iterable] = None,
    check_nans: bool = True,
    **trace_kwargs,
):
    """Add a trace to the figure.

    Parameters
    ----------
    trace : BaseTraceType or dict
        Either:

          - An instances of a trace class from the ``plotly.graph_objects`` (go)
            package (e.g., ``go.Scatter``, ``go.Bar``)
          - or a dict where:

            - The type property specifies the trace type (e.g. scatter, bar,
              area, etc.). If the dict has no 'type' property then scatter is
              assumed.
            - All remaining properties are passed to the constructor
              of the specified trace type.
    max_n_samples : int, optional
        The maximum number of samples that will be shown by the trace.\n
        !!! note
            If this variable is not set; ``_global_n_shown_samples`` will be used.
    downsampler: AbstractAggregator, optional
        The abstract series downsampler method.\n
        !!! note
            If this variable is not set, ``_global_downsampler`` will be used.
    gap_handler: AbstractGapHandler, optional
        The abstract series gap handler method.\n
        !!! note
            If this variable is not set, ``_global_gap_handler`` will be used.
    limit_to_view: boolean, optional
        If set to True, the trace's datapoints will be cut to the corresponding
        front-end view, even if the total number of samples is lower than
        ``max_n_samples``, By default False.\n
        Remark that setting this parameter to True ensures that low frequency traces
        are added to the ``hf_data`` property.
    hf_x: Iterable, optional
        The original high frequency series positions, can be either a time-series or
        an increasing, numerical index. If set, this has priority over the trace its
        data.
    hf_y: Iterable, optional
        The original high frequency values. If set, this has priority over the
        trace its data.
    hf_text: Iterable, optional
        The original high frequency text. If set, this has priority over the trace
        its ``text`` argument.
    hf_hovertext: Iterable, optional
        The original high frequency hovertext. If set, this has priority over the
        trace its ```hovertext`` argument.
    hf_marker_size: Iterable, optional
        The original high frequency marker size. If set, this has priority over the
        trace its ``marker.size`` argument.
    hf_marker_color: Iterable, optional
        The original high frequency marker color. If set, this has priority over the
        trace its ``marker.color`` argument.
    check_nans: boolean, optional
        If set to True, the trace's data will be checked for NaNs - which will be
        removed. By default True.
        As this is a costly operation, it is recommended to set this parameter to
        False if you are sure that your data does not contain NaNs (or when the
        downsampler can handle NaNs, e.g., EveryNthPoint). This should considerably
        speed up the graph construction time.
    **trace_kwargs: dict
        Additional trace related keyword arguments.
        e.g.: row=.., col=..., secondary_y=...

        !!! info "See Also"
            [`Figure.add_trace`](https://plotly.com/python-api-reference/generated/plotly.graph_objects.Figure.html#plotly.graph_objects.Figure.add_trace>) docs.

    Returns
    -------
    BaseFigure
        The Figure on which ``add_trace`` was called on; i.e. self.

    !!! note

        Constructing traces with **very large data amounts** really takes some time.
        To speed this up; use this [`add_trace`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_trace] method and

        1. Create a trace with no data (empty lists)
        2. pass the high frequency data to this method using the ``hf_x`` and ``hf_y``
           parameters.

        See the example below:
            ```python
            >>> from plotly.subplots import make_subplots
            >>> s = pd.Series()  # a high-frequency series, with more than 1e7 samples
            >>> fig = FigureResampler(go.Figure())
            >>> fig.add_trace(go.Scattergl(x=[], y=[], ...), hf_x=s.index, hf_y=s)
            ```

        !!! todo
            * explain why adding x and y to a trace is so slow
            * check and simplify the example above

    !!! tip

        * If you **do not want to downsample** your data, set ``max_n_samples`` to the
          the number of datapoints of your trace!

    !!! warning

        * The ``NaN`` values in either ``hf_y`` or ``trace.y`` will be omitted! We do
          not allow ``NaN`` values in ``hf_x`` or ``trace.x``.
        * ``hf_x``, ``hf_y``, ``hf_text``, and ``hf_hovertext`` are useful when you deal
          with large amounts of data (as it can increase the speed of this add_trace()
          method with ~30%). These arguments have priority over the trace's data and
          (hover)text attributes.
        * Low-frequency time-series data, i.e. traces that are not resampled, can hinder
          the the automatic-zooming (y-scaling) as these will not be stored in the
          back-end and thus not be scaled to the view.
          To circumvent this, the ``limit_to_view`` argument can be set, resulting in
          also storing the low-frequency series in the back-end.

    """
    # to comply with the plotly data input acceptance behavior
    if isinstance(trace, (list, tuple)):
        raise ValueError("Trace must be either a dict or a BaseTraceType")

    max_out_s = (
        self._global_n_shown_samples if max_n_samples is None else max_n_samples
    )

    # Validate the trace and convert to a trace object
    if not isinstance(trace, BaseTraceType):
        trace = self._data_validator.validate_coerce(trace)[0]

    # First add a UUID, as each (even the non-hf_data traces), must contain this
    # key for comparison. If the trace already has a UUID, we will keep it.
    uuid_str = str(uuid4()) if trace.uid is None else trace.uid
    trace.uid = uuid_str

    # construct the hf_data_container
    # TODO in future version -> maybe regex on kwargs which start with `hf_`
    dc = self._parse_get_trace_props(
        trace,
        hf_x,
        hf_y,
        hf_text,
        hf_hovertext,
        hf_marker_size,
        hf_marker_color,
        check_nans,
    )

    # These traces will determine the autoscale its RANGE!
    #   -> so also store when `limit_to_view` is set.
    if trace["type"].lower() in self._high_frequency_traces:
        n_samples = len(dc.x)
        if n_samples > max_out_s or limit_to_view:
            self._print(
                f"\t[i] DOWNSAMPLE {trace['name']}\t{n_samples}->{max_out_s}"
            )

            self._hf_data[uuid_str] = self._construct_hf_data_dict(
                dc,
                trace=trace,
                downsampler=downsampler,
                gap_handler=gap_handler,
                max_n_samples=max_n_samples,
            )

            # Before we update the trace, we create a new pointer to that trace in
            # which the downsampled data will be stored. This way, the original
            # data of the trace to this `add_trace` method will not be altered.
            # We copy (by reference) all the non-data properties of the trace in
            # the new trace.
            trace = trace._props  # convert the trace into a dict
            # NOTE: there is no need to store `marker` property here.
            # If needed, it will be added  to `trace` via `check_update_trace_data`
            trace = {
                k: trace[k] for k in set(trace.keys()).difference(set(dc._fields))
            }

            # NOTE:
            # If all the raw data needs to be sent to the javascript, and the trace
            # is high-frequency, this would take significant time!
            # Hence, you first downsample the trace.
            trace = self._check_update_trace_data(trace)
            assert trace is not None
            return super(AbstractFigureAggregator, self).add_traces(
                [trace], **self._add_trace_to_add_traces_kwargs(trace_kwargs)
            )
        else:
            self._print(f"[i] NOT resampling {trace['name']} - len={n_samples}")
            trace._process_kwargs(**{k: getattr(dc, k) for k in dc._fields})
            return super(AbstractFigureAggregator, self).add_traces(
                [trace], **self._add_trace_to_add_traces_kwargs(trace_kwargs)
            )

    return super(self._figure_class, self).add_traces(
        [trace], **self._add_trace_to_add_traces_kwargs(trace_kwargs)
    )

add_traces(data, max_n_samples=None, downsamplers=None, gap_handlers=None, limit_to_views=False, check_nans=True, **traces_kwargs)

Add traces to the figure.

Note

Make sure to look at the add_trace function for more info about speed optimization, and dealing with not high-frequency data, but still want to resample / limit the data to the front-end view.

Parameters:

Name	Type	Description	Default
data	List[BaseTraceType | dict]	A list of trace specifications to be added. Trace specifications may be either: Instances of trace classes from the plotly.graph_objs package (e.g plotly.graph_objs.Scatter, plotly.graph_objs.Bar). Dicts where: The ‘type’ property specifies the trace type (e.g. ‘scatter’, ‘bar’, ‘area’, etc.). If the dict has no ‘type’ property then ‘scatter’ is assumed. All remaining properties are passed to the constructor of the specified trace type.	required
max_n_samples	None | List[int] | int	The maximum number of samples that will be shown for each trace. If a single integer is passed, all traces will use this number. If this variable is not set; _global_n_shown_samples will be used.	None
downsamplers	None | List[AbstractAggregator] | AbstractAggregator	The downsampler that will be used to aggregate the traces. If a single aggregator is passed, all traces will use this aggregator. If this variable is not set, _global_downsampler will be used.	None
gap_handlers	None | List[AbstractGapHandler] | AbstractGapHandler	The gap handler that will be used to aggregate the traces. If a single gap handler is passed, all traces will use this gap handler. If this variable is not set, _global_gap_handler will be used.	None
limit_to_views	None | List[bool] | bool	List of limit_to_view booleans for the added traces. If set to True the trace’s datapoints will be cut to the corresponding front-end view, even if the total number of samples is lower than max_n_samples. If a single boolean is passed, all to be added traces will use this value, by default False. Remark that setting this parameter to True ensures that low frequency traces are added to the hf_data property.	False
check_nans	None | List[bool] | bool	List of check_nans booleans for the added traces. If set to True, the trace’s datapoints will be checked for NaNs. If a single boolean is passed, all to be added traces will use this value, by default True. As this is a costly operation, it is recommended to set this parameter to False if the data is known to contain no NaNs (or when the downsampler can handle NaNs, e.g., EveryNthPoint). This will considerably speed up the graph construction time.	True
**traces_kwargs		Additional trace related keyword arguments. e.g.: rows=.., cols=…, secondary_ys=… See Also Figure.add_traces docs.	{}

Returns:

Type	Description
BaseFigure	The Figure on which add_traces was called on; i.e. self.

Source code in plotly_resampler/figure_resampler/figure_resampler_interface.py

def add_traces(
    self,
    data: List[BaseTraceType | dict] | BaseTraceType | Dict,
    max_n_samples: None | List[int] | int = None,
    downsamplers: None | List[AbstractAggregator] | AbstractAggregator = None,
    gap_handlers: None | List[AbstractGapHandler] | AbstractGapHandler = None,
    limit_to_views: List[bool] | bool = False,
    check_nans: List[bool] | bool = True,
    **traces_kwargs,
):
    """Add traces to the figure.

    !!! note

        Make sure to look at the [`add_trace`][figure_resampler.figure_resampler_interface.AbstractFigureAggregator.add_trace] function for more info about
        **speed optimization**, and dealing with not ``high-frequency`` data, but
        still want to resample / limit the data to the front-end view.

    Parameters
    ----------
    data : List[BaseTraceType  |  dict]
        A list of trace specifications to be added.
        Trace specifications may be either:

          - Instances of trace classes from the plotly.graph_objs
            package (e.g plotly.graph_objs.Scatter, plotly.graph_objs.Bar).
          - Dicts where:

              - The 'type' property specifies the trace type (e.g.
                'scatter', 'bar', 'area', etc.). If the dict has no 'type'
                property then 'scatter' is assumed.
              - All remaining properties are passed to the constructor
                of the specified trace type.
    max_n_samples : None | List[int] | int, optional
          The maximum number of samples that will be shown for each trace.
          If a single integer is passed, all traces will use this number. If this
          variable is not set; ``_global_n_shown_samples`` will be used.
    downsamplers : None | List[AbstractAggregator] | AbstractAggregator, optional
        The downsampler that will be used to aggregate the traces. If a single
        aggregator is passed, all traces will use this aggregator.
        If this variable is not set, ``_global_downsampler`` will be used.
    gap_handlers : None | List[AbstractGapHandler] | AbstractGapHandler, optional
        The gap handler that will be used to aggregate the traces. If a single
        gap handler is passed, all traces will use this gap handler.
        If this variable is not set, ``_global_gap_handler`` will be used.
    limit_to_views : None | List[bool] | bool, optional
        List of limit_to_view booleans for the added traces. If set to True the
        trace's datapoints will be cut to the corresponding front-end view, even if
        the total number of samples is lower than ``max_n_samples``.
        If a single boolean is passed, all to be added traces will use this value,
        by default False.\n
        Remark that setting this parameter to True ensures that low frequency traces
        are added to the ``hf_data`` property.
    check_nans : None | List[bool] | bool, optional
        List of check_nans booleans for the added traces. If set to True, the
        trace's datapoints will be checked for NaNs. If a single boolean is passed,
        all to be added traces will use this value, by default True.\n
        As this is a costly operation, it is recommended to set this parameter to
        False if the data is known to contain no NaNs (or when the downsampler can
        handle NaNs, e.g., EveryNthPoint). This will considerably speed up the graph
        construction time.
    **traces_kwargs: dict
        Additional trace related keyword arguments.
        e.g.: rows=.., cols=..., secondary_ys=...

        !!! info "See Also"

            [`Figure.add_traces`](https://plotly.com/python-api-reference/generated/plotly.graph_objects.Figure.html#plotly.graph_objects.Figure.add_traces>) docs.

    Returns
    -------
    BaseFigure
        The Figure on which ``add_traces`` was called on; i.e. self.

    """
    # note: Plotly its add_traces also a allows non list-like input e.g. a scatter
    # object; the code below is an exact copy of their internally applied parsing
    if not isinstance(data, (list, tuple)):
        data = [data]

    # Convert each trace into a BaseTraceType object
    data = [
        self._data_validator.validate_coerce(trace)[0]
        if not isinstance(trace, BaseTraceType)
        else trace
        for trace in data
    ]

    # First add a UUID, as each (even the non-hf_data traces), must contain this
    # key for comparison. If the trace already has a UUID, we will keep it.
    for trace in data:
        uuid_str = str(uuid4()) if trace.uid is None else trace.uid
        trace.uid = uuid_str

    # Convert the data properties
    if isinstance(max_n_samples, (int, np.integer)) or max_n_samples is None:
        max_n_samples = [max_n_samples] * len(data)
    if isinstance(downsamplers, AbstractAggregator) or downsamplers is None:
        downsamplers = [downsamplers] * len(data)
    if isinstance(gap_handlers, AbstractGapHandler) or gap_handlers is None:
        gap_handlers = [gap_handlers] * len(data)
    if isinstance(limit_to_views, bool):
        limit_to_views = [limit_to_views] * len(data)
    if isinstance(check_nans, bool):
        check_nans = [check_nans] * len(data)

    zipped = zip(
        data, max_n_samples, downsamplers, gap_handlers, limit_to_views, check_nans
    )
    for (
        i,
        (trace, max_out, downsampler, gap_handler, limit_to_view, check_nan),
    ) in enumerate(zipped):
        if (
            trace.type.lower() not in self._high_frequency_traces
            or self._hf_data.get(trace.uid) is not None
        ):
            continue

        max_out_s = self._global_n_shown_samples if max_out is None else max_out
        if not limit_to_view and (trace.y is None or len(trace.y) <= max_out_s):
            continue

        dc = self._parse_get_trace_props(trace, check_nans=check_nan)
        self._hf_data[trace.uid] = self._construct_hf_data_dict(
            dc,
            trace=trace,
            downsampler=downsampler,
            gap_handler=gap_handler,
            max_n_samples=max_out,
            offset=i,
        )

        # convert the trace into a dict, and only withholds the non-hf props
        trace = trace._props
        trace = {k: trace[k] for k in set(trace.keys()).difference(set(dc._fields))}

        # update the trace data with the HF props
        trace = self._check_update_trace_data(trace)
        assert trace is not None
        data[i] = trace

    return super(self._figure_class, self).add_traces(data, **traces_kwargs)

construct_update_data(relayout_data)

Construct the to-be-updated front-end data, based on the layout change.

Attention

This method is tightly coupled with Dash app callbacks. It takes the front-end figure its relayoutData as input and returns the data which needs to be sent tot the TraceUpdater its updateData property for that corresponding graph.

Parameters:

Name	Type	Description	Default
relayout_data	dict	A dict containing the relayout-data (a.k.a. changed layout data) of the corresponding front-end graph.	required

Returns:

Type	Description
List[dict]:	A list of dicts, where each dict-item is a representation of a trace its data properties which are affected by the front-end layout change. |br| In other words, only traces which need to be updated will be sent to the front-end. Additionally, each trace-dict withholds the index of its corresponding position in the figure[data] array with the index-key in each dict.

Source code in plotly_resampler/figure_resampler/figure_resampler_interface.py

def construct_update_data(
    self,
    relayout_data: dict,
) -> Union[List[dict], dash.no_update]:
    """Construct the to-be-updated front-end data, based on the layout change.

    Attention
    ---------
    This method is tightly coupled with Dash app callbacks. It takes the front-end
    figure its ``relayoutData`` as input and returns the data which needs to be
    sent tot the ``TraceUpdater`` its ``updateData`` property for that corresponding
    graph.

    Parameters
    ----------
    relayout_data: dict
        A dict containing the ``relayout``-data (a.k.a. changed layout data) of
        the corresponding front-end graph.

    Returns
    -------
    List[dict]:
        A list of dicts, where each dict-item is a representation of a trace its
        *data* properties which are affected by the front-end layout change. |br|
        In other words, only traces which need to be updated will be sent to the
        front-end. Additionally, each trace-dict withholds the *index* of its
        corresponding position in the ``figure[data]`` array with the ``index``-key
        in each dict.

    """
    current_graph = self._get_current_graph()
    updated_trace_indices, cl_k = [], []
    if relayout_data:
        self._print("-" * 100 + "\n", "changed layout", relayout_data)

        cl_k = list(relayout_data.keys())

        # ------------------ HF DATA aggregation ---------------------
        # 1. Base case - there is an x-range specified in the front-end
        start_matches = self._re_matches(re.compile(r"xaxis\d*.range\[0]"), cl_k)
        stop_matches = self._re_matches(re.compile(r"xaxis\d*.range\[1]"), cl_k)
        if start_matches and stop_matches:  # when both are not empty
            for t_start_key, t_stop_key in zip(start_matches, stop_matches):
                # Check if the xaxis<NUMB> part of xaxis<NUMB>.[0-1] matches
                xaxis = t_start_key.split(".")[0]
                assert xaxis == t_stop_key.split(".")[0]
                # -> we want to copy the layout on the back-end
                updated_trace_indices = self._check_update_figure_dict(
                    current_graph,
                    start=relayout_data[t_start_key],
                    stop=relayout_data[t_stop_key],
                    xaxis_filter=xaxis,
                    updated_trace_indices=updated_trace_indices,
                )

        # 2. The user clicked on either autorange | reset axes
        autorange_matches = self._re_matches(
            re.compile(r"xaxis\d*.autorange"), cl_k
        )
        spike_matches = self._re_matches(re.compile(r"xaxis\d*.showspikes"), cl_k)
        # 2.1 Reset-axes -> autorange & reset to the global data view
        if autorange_matches and spike_matches:  # when both are not empty
            for autorange_key in autorange_matches:
                if relayout_data[autorange_key]:
                    xaxis = autorange_key.split(".")[0]
                    updated_trace_indices = self._check_update_figure_dict(
                        current_graph,
                        xaxis_filter=xaxis,
                        updated_trace_indices=updated_trace_indices,
                    )
        # 2.1. Autorange -> do nothing, the autorange will be applied on the
        #      current front-end view
        elif (
            autorange_matches and not spike_matches
        ):  # when only autorange is not empty
            # PreventUpdate returns a 204 status code response on the
            # relayout post request
            return dash.no_update

    # If we do not have any traces to be updated, we will return an empty
    # request response
    if not updated_trace_indices:  # when updated_trace_indices is empty
        # PreventUpdate returns a 204 status-code response on the relayout post
        # request
        return dash.no_update

    # -------------------- construct callback data --------------------------
    # 1. Create the layout data for the front-end
    layout_traces_list: List[dict] = [relayout_data]

    # 2. Create the additional trace data for the frond-end
    relevant_keys = list(_hf_data_container._fields) + ["name", "marker"]
    # Note that only updated trace-data will be sent to the client
    for idx in updated_trace_indices:
        trace = current_graph["data"][idx]
        # TODO: check if we can reduce even more
        trace_reduced = {k: trace[k] for k in relevant_keys if k in trace}

        # Store the index into the corresponding to-be-sent trace-data so
        # the client front-end can know which trace needs to be updated
        trace_reduced.update({"index": idx})
        layout_traces_list.append(trace_reduced)
    return layout_traces_list

replace(figure, convert_existing_traces=True)

Replace the current figure layout with the passed figure object.

Parameters:

Name	Type	Description	Default
figure	go.Figure	The figure object which will replace the existing figure.	required
convert_existing_traces	bool	A bool indicating whether the traces of the passed figure should be resampled, by default True.	True

Source code in plotly_resampler/figure_resampler/figure_resampler_interface.py

def replace(self, figure: go.Figure, convert_existing_traces: bool = True):
    """Replace the current figure layout with the passed figure object.

    Parameters
    ----------
    figure: go.Figure
        The figure object which will replace the existing figure.
    convert_existing_traces: bool, Optional
        A bool indicating whether the traces of the passed ``figure`` should be
        resampled, by default True.

    """
    self._clear_figure()
    self.__init__(
        figure=figure,
        convert_existing_traces=convert_existing_traces,
        default_n_shown_samples=self._global_n_shown_samples,
        default_downsampler=self._global_downsampler,
        default_gap_handler=self._global_gap_handler,
        resampled_trace_prefix_suffix=(self._prefix, self._suffix),
        show_mean_aggregation_size=self._show_mean_aggregation_size,
        verbose=self._print_verbose,
    )