Source code for fsleyes.views.plotpanel

#
# plotpanel.py - The PlotPanel class.
#
# Author: Paul McCarthy <pauldmccarthy@gmail.com>
#
"""This module provides the :class:`PlotPanel` and :class:`.OverlayPlotPanel`
classes.  The ``PlotPanel`` class is the base class for all *FSLeyes views*
which display some sort of data plot. The ``OverlayPlotPanel`` is a
``PlotPanel`` which contains some extra logic for displaying plots related to
the currently selected overlay.
"""


import logging
import collections

import wx

import numpy             as np
import scipy.interpolate as interp

import matplotlib.pyplot as plt
from matplotlib.backends.backend_wxagg import FigureCanvasWxAgg as Canvas

import fsl.utils.idle                     as idle
import fsleyes_props                      as props
import fsleyes_widgets                    as fwidgets
import fsleyes_widgets.elistbox           as elistbox

import fsleyes.strings                    as strings
import fsleyes.actions                    as actions
import fsleyes.overlay                    as fsloverlay
import fsleyes.colourmaps                 as fslcm
import fsleyes.plotting                   as plotting
import fsleyes.controls.overlaylistpanel  as overlaylistpanel
import fsleyes.controls.plotlistpanel     as plotlistpanel
from . import                                viewpanel


log = logging.getLogger(__name__)


[docs]class PlotPanel(viewpanel.ViewPanel): """The ``PlotPanel`` class is the base class for all *FSLeyes views* which display some sort of 2D data plot, such as the :class:`.TimeSeriesPanel`, and the :class:`.HistogramPanel`. See also the :class:`OverlayPlotPanel`, which contains extra logic for displaying plots related to the currently selected overlay. ``PlotPanel`` uses :mod:`matplotlib` for its plotting. The ``matplotlib`` ``Figure``, ``Axis``, and ``Canvas`` instances can be accessed via the :meth:`getFigure`, :meth:`getAxis`, and :meth:`getCanvas` methods, if they are needed. Various display settings can be configured through ``PlotPanel`` properties, including :attr:`legend`, :attr:`smooth`, etc. **Sub-class requirements** Sub-class implementations of ``PlotPanel`` must do the following: 1. Call the ``PlotPanel`` constructor. 2. Define a :class:`.DataSeries` sub-class. 3. Override the :meth:`draw` method, so it calls the :meth:`drawDataSeries` method. 4. If necessary, override the :meth:`prepareDataSeries` method to perform any preprocessing on ``extraSeries`` passed to the :meth:`drawDataSeries` method (but not applied to :class:`.DataSeries` that have been added to the :attr:`dataSeries` list). 5. If necessary, override the :meth:`destroy` method, but make sure that the base-class implementation is called. **Data series** A ``PlotPanel`` instance plots data contained in one or more :class:`.DataSeries` instances; all ``DataSeries`` classes are defined in the :mod:`.plotting` sub-package. Therefore, ``PlotPanel`` sub-classes also need to define a sub-class of the :class:`.DataSeries` base class. ``DataSeries`` objects can be plotted by passing them to the :meth:`drawDataSeries` method. Or, if you want one or more ``DataSeries`` to be *held*, i.e. plotted every time, you can add them to the :attr:`dataSeries` list. The ``DataSeries`` in the :attr:`dataSeries` list will be plotted on every call to :meth:`drawDataSeries` (in addition to any ``DataSeries`` passed directly to :meth:`drawDataSeries`) until they are removed from the :attr:`dataSeries` list. **The draw queue** The ``PlotPanel`` uses a :class:`.async.TaskThread` to asynchronously extract and prepare data for plotting, This is because data preparation may take a long time for large :class:`.Image` overlays, and the main application thread should not be blocked while this is occurring. The ``TaskThread`` instance is accessible through the :meth:`getDrawQueue` method, in case anything needs to be scheduled on it. **Plot panel actions** A number of :mod:`actions` are also provided by the ``PlotPanel`` class: .. autosummary:: :nosignatures: screenshot importDataSeries exportDataSeries """ dataSeries = props.List() """This list contains :class:`.DataSeries` instances which are plotted on every call to :meth:`drawDataSeries`. ``DataSeries`` instances can be added/removed directly to/from this list. """ artists = props.List() """This list contains any ``matplotlib.Artist`` instances which are plotted every call to :meth:`drawArtists`. """ legend = props.Boolean(default=True) """If ``True``, a legend is added to the plot, with an entry for every ``DataSeries`` instance in the :attr:`dataSeries` list. """ xAutoScale = props.Boolean(default=True) """If ``True``, the plot :attr:`limits` for the X axis are automatically updated to fit all plotted data. """ yAutoScale = props.Boolean(default=True) """If ``True``, the plot :attr:`limits` for the Y axis are automatically updated to fit all plotted data. """ xLogScale = props.Boolean(default=False) """Toggle a :math:`log_{10}` x axis scale. """ yLogScale = props.Boolean(default=False) """Toggle a :math:`log_{10}` y axis scale. """ invertX = props.Boolean(default=False) """Invert the plot along the X axis. """ invertY = props.Boolean(default=False) """Invert the plot along the Y axis. """ xScale = props.Real(default=1) """Scale to apply to the X axis data. """ yScale = props.Real(default=1) """Scale to apply to the Y axis data. """ xOffset = props.Real(default=0) """Offset to apply to the X axis data. """ yOffset = props.Real(default=0) """Offset to apply to the Y axis data. """ ticks = props.Boolean(default=True) """Toggle axis ticks and tick labels on/off.""" grid = props.Boolean(default=True) """Toggle an axis grid on/off.""" gridColour = props.Colour(default=(1, 1, 1)) """Grid colour (if :attr:`grid` is ``True``).""" bgColour = props.Colour(default=(0.8, 0.8, 0.8)) """Plot background colour.""" smooth = props.Boolean(default=False) """If ``True`` all plotted data is up-sampled, and smoothed using spline interpolation. """ xlabel = props.String() """A label to show on the x axis. """ ylabel = props.String() """A label to show on the y axis. """ limits = props.Bounds(ndims=2) """The x/y axis limits. If :attr:`xAutoScale` and :attr:`yAutoScale` are ``True``, these limit values are automatically updated on every call to :meth:`drawDataSeries`. """
[docs] def __init__(self, parent, overlayList, displayCtx, frame): """Create a ``PlotPanel``. :arg parent: The :mod:`wx` parent object. :arg overlayList: An :class:`.OverlayList` instance. :arg displayCtx: A :class:`.DisplayContext` instance. :arg frame: The :class:`.FSLeyesFrame` instance. """ viewpanel.ViewPanel.__init__( self, parent, overlayList, displayCtx, frame) figure = plt.Figure() axis = figure.add_subplot(111) canvas = Canvas(self, -1, figure) figure.subplots_adjust(top=1.0, bottom=0.0, left=0.0, right=1.0) figure.patch.set_visible(False) self.centrePanel = canvas self.__figure = figure self.__axis = axis self.__canvas = canvas self.__name = 'PlotPanel_{}'.format(self.name) # Accessing data from large compressed # files may take time, so we maintain # a queue of plotting requests. The # functions executed on this task # thread are used to prepare data for # plotting - the plotting occurs on # the main WX event loop. # # The drawDataSeries method sets up the # asynchronous data preparation, and the # __drawDataSeries method does the actual # plotting. self.__drawQueue = idle.TaskThread() self.__drawQueue.daemon = True self.__drawQueue.start() # Whenever a new request comes in to # draw the plot, we can't cancel any # pending requests, as they are running # on separate threads and out of our # control (and could be blocking on I/O). # # Instead, we keep track of the total # number of pending requests. The # __drawDataSeries method (which does the # actual plotting) will only draw the # plot if there are no pending requests # (because otherwise it would be drawing # out-of-date data). self.__drawRequests = 0 # The getDrawnDataSeries method returns # data as it is shown on the plot - some # pre/post-processing may be applied to # the data as retrieved by DataSeries # instances, so this dictionary is used # to keep copies of the mpl Artist object # which contains the data with all # processing applied, that is currrently # on the plot (and accessible via # getDrawnDataSeries). self.__drawnDataSeries = collections.OrderedDict() # Redraw whenever any property changes, for propName in ['legend', 'xAutoScale', 'yAutoScale', 'xLogScale', 'yLogScale', 'invertX', 'invertY', 'xScale', 'yScale', 'xOffset', 'yOffset', 'ticks', 'grid', 'gridColour', 'bgColour', 'smooth', 'xlabel', 'ylabel']: self.addListener(propName, self.__name, self.asyncDraw) # custom listeners for a couple of properties self.addListener('dataSeries', self.__name, self.__dataSeriesChanged) self.addListener('artists', self.__name, self.__artistsChanged) self.addListener('limits', self.__name, self.__limitsChanged)
[docs] def getFigure(self): """Returns the ``matplotlib`` ``Figure`` instance.""" return self.__figure
[docs] def getAxis(self): """Returns the ``matplotlib`` ``Axis`` instance.""" return self.__axis
[docs] def getCanvas(self): """Returns the ``matplotlib`` ``Canvas`` instance.""" return self.__canvas
[docs] def getDrawQueue(self): """Returns the :class`.idle.TaskThread` instance used for data preparation. """ return self.__drawQueue
[docs] def draw(self, *a): """This method must be overridden by ``PlotPanel`` sub-classes. It is called whenever a :class:`.DataSeries` is added to the :attr:`dataSeries` list, or when any plot display properties change. Sub-class implementations should call the :meth:`drawDataSeries` and meth:`drawArtists` methods. """ raise NotImplementedError('The draw method must be ' 'implemented by PlotPanel subclasses')
[docs] def asyncDraw(self, *a): """Schedules :meth:`draw` to be run asynchronously. This method should be used in preference to calling :meth:`draw` directly in most cases, particularly where the call occurs within a property callback function. """ idleName = '{}.draw'.format(id(self)) if not self.destroyed and not idle.idleLoop.inIdle(idleName): idle.idle(self.draw, name=idleName)
[docs] def destroy(self): """Removes some property listeners, and then calls :meth:`.ViewPanel.destroy`. """ self.removeListener('dataSeries', self.__name) self.removeListener('artists', self.__name) self.removeListener('limits', self.__name) for propName in ['legend', 'xAutoScale', 'yAutoScale', 'xLogScale', 'yLogScale', 'invertX', 'invertY', 'xScale', 'yScale', 'xOffset', 'yOffset', 'ticks', 'grid', 'gridColour', 'bgColour', 'smooth', 'xlabel', 'ylabel']: self.removeListener(propName, self.__name) for ds in self.dataSeries: for propName in ds.redrawProperties(): ds.removeListener(propName, self.__name) ds.destroy() self.__drawQueue.stop() self.__drawQueue = None self.__drawnDataSeries = None self.dataSeries = [] self.artists = [] self.__figure = None self.__axis = None self.__canvas = None viewpanel.ViewPanel.destroy(self)
[docs] @actions.action def screenshot(self, *a): """Prompts the user to select a file name, then saves a screenshot of the current plot. See the :class:`.ScreenshotAction`. """ from fsleyes.actions.screenshot import ScreenshotAction ScreenshotAction(self.overlayList, self.displayCtx, self)()
[docs] @actions.action def importDataSeries(self, *a): """Imports data series from a text file. See the :class:`.ImportDataSeriesAction`. """ from fsleyes.actions.importdataseries import ImportDataSeriesAction ImportDataSeriesAction(self.overlayList, self.displayCtx, self)()
[docs] @actions.action def exportDataSeries(self, *args, **kwargs): """Exports displayed data series to a text file. See the :class:`.ExportDataSeriesAction`. """ from fsleyes.actions.exportdataseries import ExportDataSeriesAction ExportDataSeriesAction(self.overlayList, self.displayCtx, self)()
[docs] def message(self, msg, clear=True, border=False): """Displays the given message in the centre of the figure. This is a convenience method provided for use by subclasses. """ axis = self.getAxis() if clear: self.__drawnDataSeries.clear() axis.clear() axis.set_xlim((0.0, 1.0)) axis.set_ylim((0.0, 1.0)) if border: bbox = {'facecolor' : '#ffffff', 'edgecolor' : '#cdcdff', 'boxstyle' : 'round,pad=1'} else: bbox = None axis.text(0.5, 0.5, msg, ha='center', va='center', transform=axis.transAxes, bbox=bbox) self.getCanvas().draw() self.Refresh()
[docs] def getArtist(self, ds): """Returns the ``matplotlib.Artist`` (typically a ``Line2D`` instance) associated with the given :class:`.DataSeries` instance. A ``KeyError`` is raised if there is no such artist. """ return self.__drawnDataSeries[ds]
[docs] def getDrawnDataSeries(self): """Returns a list of tuples, each tuple containing the ``(DataSeries, x, y)`` data for one ``DataSeries`` instance as it is shown on the plot. """ return [(ds, np.array(l.get_xdata()), np.array(l.get_ydata())) for ds, l in self.__drawnDataSeries.items()]
[docs] def prepareDataSeries(self, ds): """Prepares the data from the given :class:`.DataSeries` so it is ready to be plotted. Called by the :meth:`__drawOneDataSeries` method for any ``extraSeries`` passed to the :meth:`drawDataSeries` method (but not applied to :class:`.DataSeries` that have been added to the :attr:`dataSeries` list). This implementation just returns :class:`.DataSeries.getData` - override it to perform any custom preprocessing. """ return ds.getData()
[docs] def drawArtists(self, refresh=True, immediate=False): """Draw all ``matplotlib.Artist`` instances in the :attr:`artists` list, then refresh the canvas. :arg refresh: If ``True`` (default), the canvas is refreshed. """ axis = self.getAxis() canvas = self.getCanvas() def realDraw(): # Just in case this PlotPanel is destroyed # before this task gets executed if not fwidgets.isalive(self): return for artist in self.artists: if artist not in axis.findobj(type(artist)): axis.add_artist(artist) if immediate: realDraw() else: self.__drawQueue.enqueue(idle.idle, realDraw) def refreshCanvas(): if not self.destroyed: canvas.draw() if refresh: if immediate: refreshCanvas() else: self.__drawQueue.enqueue(idle.idle, refreshCanvas)
[docs] def drawDataSeries(self, extraSeries=None, refresh=False, **plotArgs): """Queues a request to plot all of the :class:`.DataSeries` instances in the :attr:`dataSeries` list. This method does not do the actual plotting - it is performed asynchronously, to avoid locking up the GUI: 1. The data for each ``DataSeries`` instance is prepared on separate threads (using :func:`.idle.run`). 2. A call to :func:`.idle.wait` is enqueued on a :class:`.TaskThread`. 3. This ``wait`` function waits until all of the data preparation threads have completed, and then passes all of the data to the :meth:`__drawDataSeries` method. :arg extraSeries: A sequence of additional ``DataSeries`` to be plotted. These series are passed through the :meth:`prepareDataSeries` method before being plotted. :arg refresh: If ``True``, the canvas is refreshed. Otherwise, you must call ``getCanvas().draw()`` manually. Defaults to ``False`` - the :meth:`drawArtists` method will refresh the canvas, so if you call :meth:`drawArtists` immediately after calling this method (which you should), then you don't need to manually refresh the canvas. :arg plotArgs: Passed through to the :meth:`__drawDataSeries` method. .. note:: This method must only be called from the main application thread (the ``wx`` event loop). """ if extraSeries is None: extraSeries = [] canvas = self.getCanvas() axis = self.getAxis() toPlot = self.dataSeries[:] toPlot = [ds for ds in toPlot if ds.enabled] extraSeries = [ds for ds in extraSeries if ds.enabled] toPlot = extraSeries + toPlot preprocs = [True] * len(extraSeries) + [False] * len(toPlot) if len(toPlot) == 0: self.__drawnDataSeries.clear() axis.clear() canvas.draw() self.Refresh() return # Before clearing/redrawing, save # a copy of the x/y axis limits - # the user may have changed them # via panning/zooming and, if # autoLimit is off, we will want # to preserve the limits that the # user set. These are passed to # the __drawDataSeries method. # # Make sure the limits are ordered # as (min, max), as they won't be # if invertX/invertY are active. axxlim = list(sorted(axis.get_xlim())) axylim = list(sorted(axis.get_ylim())) # Here we are preparing the data for # each data series on separate threads, # as data preparation can be time # consuming for large images. We # display a message on the canvas # during preparation. tasks = [] allXdata = [None] * len(toPlot) allYdata = [None] * len(toPlot) # Create a separate function # for each data series for idx, (ds, preproc) in enumerate(zip(toPlot, preprocs)): def getData(d=ds, p=preproc, i=idx): if not d.enabled: return if p: xdata, ydata = self.prepareDataSeries(d) else: xdata, ydata = d.getData() allXdata[i] = xdata allYdata[i] = ydata tasks.append(getData) # Run the data preparation tasks, # a separate thread for each. tasks = [idle.run(t) for t in tasks] # Show a message while we're # preparing the data. self.message(strings.messages[self, 'preparingData'], clear=False, border=True) # Wait until data preparation is # done, then call __drawDataSeries. self.__drawRequests += 1 self.__drawQueue.enqueue(idle.wait, tasks, self.__drawDataSeries, toPlot, allXdata, allYdata, axxlim, axylim, refresh, taskName='{}.wait'.format(id(self)), wait_direct=True, **plotArgs)
def __drawDataSeries( self, dataSeries, allXdata, allYdata, oldxlim, oldylim, refresh, xlabel=None, ylabel=None, **plotArgs): """Called by :meth:`__drawDataSeries`. Plots all of the data associated with the given ``dataSeries``. :arg dataSeries: The list of :class:`.DataSeries` instances to plot. :arg allXdata: A list of arrays containing X axis data, one for each ``DataSeries``. :arg allYdata: A list of arrays containing Y axis data, one for each ``DataSeries``. :arg oldxlim: X plot limits from the previous draw. If ``xAutoScale`` is disabled, this limit is preserved. :arg oldylim: Y plot limits from the previous draw. If ``yAutoScale`` is disabled, this limit is preserved. :arg refresh: Refresh the canvas - see :meth:`drawDataSeries`. :arg xlabel: If provided, overrides the value of the :attr:`xlabel` property. :arg ylabel: If provided, overrides the value of the :attr:`ylabel` property. :arg plotArgs: Remaining arguments passed to the :meth:`__drawOneDataSeries` method. """ # Avoid spursious post-destruction # notifications (occur sporadically # during testing) if self.destroyed: return # Only draw the plot if there are no # pending draw requests. Otherwise # we would be drawing out-of-date data. self.__drawRequests -= 1 if self.__drawRequests != 0: return axis = self.getAxis() canvas = self.getCanvas() width, height = canvas.get_width_height() self.__drawnDataSeries.clear() axis.clear() xlims = [] ylims = [] for ds, xdata, ydata in zip(dataSeries, allXdata, allYdata): if any((ds is None, xdata is None, ydata is None)): continue if not ds.enabled: continue xdata = self.xOffset + self.xScale * xdata ydata = self.yOffset + self.yScale * ydata xlim, ylim = self.__drawOneDataSeries(ds, xdata, ydata, **plotArgs) if np.any(np.isclose([xlim[0], ylim[0]], [xlim[1], ylim[1]])): continue xlims.append(xlim) ylims.append(ylim) if len(xlims) == 0: xmin, xmax = 0.0, 0.0 ymin, ymax = 0.0, 0.0 else: (xmin, xmax), (ymin, ymax) = self.__calcLimits( xlims, ylims, oldxlim, oldylim, width, height) # x/y axis labels if xlabel is None: xlabel = self.xlabel if ylabel is None: ylabel = self.ylabel if xlabel is None: xlabel = '' if ylabel is None: ylabel = '' xlabel = xlabel.strip() ylabel = ylabel.strip() if xlabel != '': axis.set_xlabel(xlabel, va='bottom') axis.xaxis.set_label_coords(0.5, 10.0 / height) if ylabel != '': axis.set_ylabel(ylabel, va='top') axis.yaxis.set_label_coords(10.0 / width, 0.5) # Ticks if self.ticks: axis.tick_params(direction='in', pad=-5) axis.tick_params(axis='both', which='both', length=3) for ytl in axis.yaxis.get_ticklabels(): ytl.set_horizontalalignment('left') for xtl in axis.xaxis.get_ticklabels(): xtl.set_verticalalignment('bottom') else: # we clear the labels, but # leave the ticks, so the # axis grid gets drawn xlabels = ['' for i in range(len(axis.xaxis.get_ticklabels()))] ylabels = ['' for i in range(len(axis.yaxis.get_ticklabels()))] axis.set_xticklabels(xlabels) axis.set_yticklabels(ylabels) axis.tick_params(axis='both', which='both', length=0) # Limits if xmin != xmax: if self.invertX: axis.set_xlim((xmax, xmin)) else: axis.set_xlim((xmin, xmax)) if self.invertY: axis.set_ylim((ymax, ymin)) else: axis.set_ylim((ymin, ymax)) # legend labels = [ds.label for ds in dataSeries if ds.label is not None] if len(labels) > 0 and self.legend: handles, labels = axis.get_legend_handles_labels() legend = axis.legend( handles, labels, loc='upper right', fontsize=10, handlelength=3, fancybox=True) legend.get_frame().set_alpha(0.6) if self.grid: axis.grid(linestyle='-', color=self.gridColour, linewidth=0.5, zorder=0) else: axis.grid(False) axis.spines['right'] .set_visible(False) axis.spines['left'] .set_visible(False) axis.spines['top'] .set_visible(False) axis.spines['bottom'].set_visible(False) axis.set_axisbelow(True) axis.patch.set_facecolor(self.bgColour) self.getFigure().patch.set_alpha(0) if refresh: canvas.draw() def __drawOneDataSeries(self, ds, xdata, ydata, **plotArgs): """Plots a single :class:`.DataSeries` instance. This method is called by the :meth:`drawDataSeries` method. :arg ds: The ``DataSeries`` instance. :arg xdata: X axis data. :arg ydata: Y axis data. :arg plotArgs: May be used to customise the plot - these arguments are all passed through to the ``Axis.plot`` function. """ if ds.alpha == 0: return (0, 0), (0, 0) if len(xdata) != len(ydata) or len(xdata) == 0: log.debug('{}: data series length mismatch, or ' 'no data points (x: {}, y: {})'.format( ds.overlay.name, len(xdata), len(ydata))) return (0, 0), (0, 0) xdata = np.asarray(xdata, dtype=np.float) ydata = np.asarray(ydata, dtype=np.float) log.debug('Drawing {} for {}'.format(type(ds).__name__, ds.overlay)) # Note to self: If the smoothed data is # filled with NaNs, it is possibly due # to duplicate values in the x data, which # are not handled very well by splrep. if self.smooth: tck = interp.splrep(xdata, ydata) xdata = np.linspace(xdata[0], xdata[-1], len(xdata) * 5, dtype=np.float32) ydata = interp.splev(xdata, tck) nans = ~(np.isfinite(xdata) & np.isfinite(ydata)) xdata[nans] = np.nan ydata[nans] = np.nan if self.xLogScale: xdata[xdata <= 0] = np.nan if self.yLogScale: ydata[ydata <= 0] = np.nan if np.all(np.isnan(xdata) | np.isnan(ydata)): return (0, 0), (0, 0) kwargs = plotArgs kwargs['lw'] = kwargs.get('lw', ds.lineWidth) kwargs['alpha'] = kwargs.get('alpha', ds.alpha) kwargs['color'] = kwargs.get('color', ds.colour) kwargs['label'] = kwargs.get('label', ds.label) kwargs['ls'] = kwargs.get('ls', ds.lineStyle) axis = self.getAxis() line = axis.plot(xdata, ydata, **kwargs)[0] self.__drawnDataSeries[ds] = line if self.xLogScale: axis.set_xscale('log') posx = xdata[xdata > 0] xlimits = np.nanmin(posx), np.nanmax(posx) else: xlimits = np.nanmin(xdata), np.nanmax(xdata) if self.yLogScale: axis.set_yscale('log') posy = ydata[ydata > 0] ylimits = np.nanmin(posy), np.nanmax(posy) else: ylimits = np.nanmin(ydata), np.nanmax(ydata) return xlimits, ylimits def __dataSeriesChanged(self, *a): """Called when the :attr:`dataSeries` list changes. Adds listeners to any new :class:`.DataSeries` instances, and then calls :meth:`asyncDraw`. """ for ds in self.dataSeries: for propName in ds.redrawProperties(): ds.addListener(propName, self.__name, self.asyncDraw, overwrite=True) self.asyncDraw() def __artistsChanged(self, *a): """Called when the :attr:`artists` list changes. Calls :meth:`asyncDraw`. """ self.asyncDraw() def __limitsChanged(self, *a): """Called when the :attr:`limits` change. Updates the axis limits accordingly. """ axis = self.getAxis() axis.set_xlim(self.limits.x) axis.set_ylim(self.limits.y) self.asyncDraw() def __calcLimits(self, dataxlims, dataylims, axisxlims, axisylims, axWidth, axHeight): """Calculates and returns suitable axis limits for the current plot. Also updates the :attr:`limits` property. This method is called by the :meth:`drawDataSeries` method. If :attr:`xAutoScale` or :attr:`yAutoScale` are enabled, the limits are calculated from the data range, using the canvas width and height to maintain consistent padding around the plotted data, irrespective of the canvas size. . Otherwise, the existing axis limits are retained. :arg dataxlims: A tuple containing the (min, max) x data range. :arg dataylims: A tuple containing the (min, max) y data range. :arg axisxlims: A tuple containing the current (min, max) x axis limits. :arg axisylims: A tuple containing the current (min, max) y axis limits. :arg axWidth: Canvas width in pixels :arg axHeight: Canvas height in pixels """ if self.xAutoScale: xmin = min([lim[0] for lim in dataxlims]) xmax = max([lim[1] for lim in dataxlims]) lPad = (xmax - xmin) * (50.0 / axWidth) rPad = (xmax - xmin) * (50.0 / axWidth) xmin = xmin - lPad xmax = xmax + rPad else: xmin = axisxlims[0] xmax = axisxlims[1] if self.yAutoScale: ymin = min([lim[0] for lim in dataylims]) ymax = max([lim[1] for lim in dataylims]) bPad = (ymax - ymin) * (50.0 / axHeight) tPad = (ymax - ymin) * (50.0 / axHeight) ymin = ymin - bPad ymax = ymax + tPad else: ymin = axisylims[0] ymax = axisylims[1] self.disableListener('limits', self.__name) self.limits[:] = [xmin, xmax, ymin, ymax] self.enableListener('limits', self.__name) return (xmin, xmax), (ymin, ymax)
[docs]class OverlayPlotPanel(PlotPanel): """The ``OverlayPlotPanel`` is a :class:`.PlotPanel` which contains some extra logic for creating, storing, and drawing :class:`.DataSeries` instances for each overlay in the :class:`.OverlayList`. **Subclass requirements** Sub-classes must: 1. Implement the :meth:`createDataSeries` method, so it creates a :class:`.DataSeries` instance for a specified overlay. 2. Implement the :meth:`PlotPanel.draw` method so it calls the :meth:`.PlotPanel.drawDataSeries`, passing :class:`.DataSeries` instances for all overlays where :attr:`.Display.enabled` is ``True``. 3. Optionally implement the :meth:`prepareDataSeries` method to perform any custom preprocessing. **The internal data series store** The ``OverlayPlotPanel`` maintains a store of :class:`.DataSeries` instances, one for each compatible overlay in the :class:`.OverlayList`. The ``OverlayPlotPanel`` manages the property listeners that must be registered with each of these ``DataSeries`` to refresh the plot. These instances are created by the :meth:`createDataSeries` method, which is implemented by sub-classes. The following methods are available to sub-classes, for managing the internal store of :class:`.DataSeries` instances: .. autosummary:: :nosignatures: getDataSeries getDataSeriesToPlot clearDataSeries updateDataSeries addDataSeries removeDataSeries **Proxy images** The ``OverlayPlotPanel`` will replace all :class:`.ProxyImage` instances with their base images. This functionality was originally added to support the :attr:`.HistogramSeries.showOverlay` functionality - it adds a mask image to the :class:`.OverlayList` to display the histogram range. Sub-classes may wish to adhere to the same logic (replacing ``ProxyImage`` instances with their bases) **Control panels** The :class:`.PlotControlPanel`, :class:`.PlotListPanel`, and :class:`.OverlayListPanel` are *FSLeyes control* panels which work with the :class:`.OverlayPlotPanel`. The ``PlotControlPanel`` is not intended to be used directly - plot-specific sub-classes are used instead. The following actions can be used to toggle control panels on an ``OverlayPlotPanel``: .. autosummary:: :nosignatures: toggleOverlayList togglePlotList **Sub-classes** The ``OverlayPlotPanel`` is the base class for: .. autosummary:: :nosignatures: ~fsleyes.views.timeseriespanel.TimeSeriesPanel ~fsleyes.views.histogrampanel.HistogramPanel ~fsleyes.views.powerspectrumpanel.PowerSpectrumPanel """ plotColours = {} """This dictionary is used to store a collection of ``{overlay : colour}`` mappings. It is shared across all ``OverlayPlotPanel`` instances, so that the same (initial) colour is used for the same overlay, across multiple plots. See also :attr:`plotStyles`. Sub-classes should use the :meth:`getOverlayPlotColour` and :meth:`getOverlayPlotStyle`methods to retrieve the initial colour and linestyle to use for a given overlay. """ plotStyles = {} """This dictionary is used to store a collection of ``{overlay : colour}`` mappings - it is used in conjunction with :attr:`plotColours`. """
[docs] def __init__(self, *args, **kwargs): """Create an ``OverlayPlotPanel``. :arg initialState: Must be passed as a keyword argument. Allows you to specify the initial enabled/disabled state for each overlay. See :meth:`updateDataSeries`. If not provided, only the data series for the currently selected overlay is shown (if possible). All other argumenst are passed through to :meth:`PlotPanel.__init__`. """ initialState = kwargs.pop('initialState', None) PlotPanel.__init__(self, *args, **kwargs) self.__name = 'OverlayPlotPanel_{}'.format(self.name) # The dataSeries attribute is a dictionary of # # {overlay : DataSeries} # # mappings, containing a DataSeries instance for # each compatible overlay in the overlay list. # # refreshProps is a dict of # # {overlay : ([targets], [propNames]} # # mappings, containing the target instances and # properties which, when those properties change, # need to trigger a refresh of the plot. # # refreshCounts is a dict of: # # {target, propName : dscount} # mappings, containing all targets and # associated property names on which a listener # is currently registered, and the count of # DataSeries instances which are interested # in them. self.__dataSeries = {} self.__refreshProps = {} self.__refreshCounts = {} # Pre-generated default colours and line # styles to use - see plotColours, plotStyles, # getOverlayPlotColour, and getOverlayPlotStyle lut = fslcm.getLookupTable('paul_tol_accessible') styles = plotting.DataSeries.lineStyle.getChoices() limit = min(len(lut), len(styles)) self.__defaultColours = [l.colour for l in lut[ :limit]] self.__defaultStyles = [s for s in styles[:limit]] self .addListener('dataSeries', self.__name, self.__dataSeriesChanged) self.overlayList.addListener('overlays', self.__name, self.__overlayListChanged) self.__overlayListChanged(initialState=initialState) self.__dataSeriesChanged()
[docs] def destroy(self): """Must be called when this ``OverlayPlotPanel`` is no longer needed. Removes some property listeners, and calls :meth:`PlotPanel.destroy`. """ self.overlayList.removeListener('overlays', self.__name) self .removeListener('dataSeries', self.__name) for overlay in list(self.__dataSeries.keys()): self.clearDataSeries(overlay) self.__dataSeries = None self.__refreshProps = None self.__refreshCounts = None PlotPanel.destroy(self)
[docs] def getDataSeriesToPlot(self): """Convenience method which returns a list of overlays which have :class:`.DataSeries` that should be plotted. """ overlays = self.overlayList[:] # Display.enabled overlays = [o for o in overlays if self.displayCtx.getDisplay(o).enabled] # Replace proxy images overlays = [o.getBase() if isinstance(o, fsloverlay.ProxyImage) else o for o in overlays] # Have data series dss = [self.getDataSeries(o) for o in overlays] dss = [ds for ds in dss if ds is not None] # Gather any extra time series # associated with the base time # series objects. for i, ds in enumerate(list(reversed(dss))): extras = ds.extraSeries() dss = dss[:i + 1] + extras + dss[i + 1:] # If a base time series is disabled, # its additional ones should also # be disabled for eds in extras: eds.enabled = ds.enabled # Remove duplicates unique = [] for ds in dss: if ds not in unique: unique.append(ds) return unique
[docs] def getDataSeries(self, overlay): """Returns the :class:`.DataSeries` instance associated with the specified overlay, or ``None`` if there is no ``DataSeries`` instance. """ if isinstance(overlay, fsloverlay.ProxyImage): overlay = overlay.getBase() return self.__dataSeries.get(overlay)
[docs] def getOverlayPlotColour(self, overlay): """Returns an initial colour to use for plots associated with the given overlay. If a colour is present in the :attr:`plotColours` dictionary, it is returned. Otherwise a random colour is generated, added to ``plotColours``, and returned. """ if isinstance(overlay, fsloverlay.ProxyImage): overlay = overlay.getBase() colour = self.plotColours.get(overlay) if colour is None: idx = len(self.plotColours) % len(self.__defaultColours) colour = self.__defaultColours[idx] self.plotColours[overlay] = colour return colour
[docs] def getOverlayPlotStyle(self, overlay): """Returns an initial line style to use for plots associated with the given overlay. If a colour is present in the :attr:`plotStyles` dictionary, it is returned. Otherwise a line style is generated, added to ``plotStyles``, and returned. The format of the returned line style is suitable for use with the ``linestyle`` argument of the ``matplotlib`` ``plot``functions. """ if isinstance(overlay, fsloverlay.ProxyImage): overlay = overlay.getBase() style = self.plotStyles.get(overlay) if style is None: idx = len(self.plotStyles) % len(self.__defaultStyles) style = self.__defaultStyles[idx] self.plotStyles[overlay] = style return style
[docs] @actions.action def addDataSeries(self): """Every :class:`.DataSeries` which is currently plotted, and has not been added to the :attr:`PlotPanel.dataSeries` list, is added to said list. """ # Get all the DataSeries objects which # have been drawn, and are not in the # dataSeries list. toAdd = self.getDrawnDataSeries() toAdd = [d[0] for d in toAdd if d[0] not in self.dataSeries] if len(toAdd) == 0: return # Replace each DataSeries instance with a copy. # This is necessary because some DataSeries # sub-classes have complicated behaviour (e.g. # changing their data when some properties # change). But we just want to 'freeze' the # data as it is currently shown. So we create # a dumb copy. for i, ds in enumerate(toAdd): copy = plotting.DataSeries(ds.overlay, self.overlayList, self.displayCtx, self) toAdd[i] = copy copy.alpha = ds.alpha copy.lineWidth = ds.lineWidth copy.lineStyle = ds.lineStyle copy.label = ds.label copy.colour = ds.colour # We have to re-generate the data, # because the x/y data returned by # the getDrawnDataSeries method # above may have had post-processing # applied to it (e.g. smoothing) xdata, ydata = self.prepareDataSeries(ds) copy.setData(xdata, ydata) # This is disgraceful. It wasn't too bad # when this function was defined in the # PlotListPanel class, but is a horrendous # hack now that it is defined here in the # PlotPanel class. # # At some stage I will remove this offensive # code, and figure out a more robust system # for appending this metadata to DataSeries # instances. # # When the user selects a data series in # the list, we want to change the selected # overlay/location/volume/etc to the # properties associated with the data series. # So here we're adding some attributes to # each data series instance so that the # PlotListPanel.__onListSelect method can # update the display properties. opts = self.displayCtx.getOpts(ds.overlay) if isinstance(ds, (plotting.MelodicTimeSeries, plotting.MelodicPowerSpectrumSeries)): copy._volume = opts.volume elif isinstance(ds, plotting.VoxelDataSeries): copy._location = opts.getVoxel() self.dataSeries.extend(toAdd)
[docs] @actions.action def removeDataSeries(self, *a): """Removes the most recently added :class:`.DataSeries` from this ``OverlayPlotPanel``. """ if len(self.dataSeries) > 0: self.dataSeries.pop()
[docs] def createDataSeries(self, overlay): """This method must be implemented by sub-classes. It must create and return a :class:`.DataSeries` instance for the specified overlay. .. note:: Sub-class implementations should set the :attr:`.DataSeries.colour` property to that returned by the :meth:`getOverlayPlotColour` method, and the :attr:`.DataSeries.lineStyle` property to that returned by the :meth:`getOverlayPlotStyle` method Different ``DataSeries`` types need to be re-drawn when different properties change. For example, a :class:`.VoxelTimeSeries`` instance needs to be redrawn when the :attr:`.DisplayContext.location` property changes, whereas a :class:`.MelodicTimeSeries` instance needs to be redrawn when the :attr:`.VolumeOpts.volume` property changes. Therefore, in addition to creating and returning a ``DataSeries`` instance for the given overlay, sub-class implementations must also specify the properties which affect the state of the ``DataSeries`` instance. These must be specified as two lists: - the *targets* list, a list of objects which own the dependant properties (e.g. the :class:`.DisplayContext` or :class:`.VolumeOpts` instance). - The *properties* list, a list of names, each specifying the property on the corresponding target. This method must therefore return a tuple containing: - A :class:`.DataSeries` instance, or ``None`` if the overlay is incompatible. - A list of *target* instances. - A list of *property names*. The target and property name lists must have the same length. """ raise NotImplementedError('createDataSeries must be ' 'implemented by sub-classes')
[docs] def clearDataSeries(self, overlay): """Destroys the internally cached :class:`.DataSeries` for the given overlay. """ if isinstance(overlay, fsloverlay.ProxyImage): overlay = overlay.getBase() ds = self.__dataSeries .pop(overlay, None) targets, propNames = self.__refreshProps.pop(overlay, ([], [])) if ds is not None: log.debug('Destroying {} for {}'.format( type(ds).__name__, overlay)) for propName in ds.redrawProperties(): ds.removeListener(propName, self.__name) ds.destroy() for t, p in zip(targets, propNames): count = self.__refreshCounts[t, p] if count - 1 == 0: self.__refreshCounts.pop((t, p)) t.removeListener(p, self.__name) else: self.__refreshCounts[t, p] = count - 1
[docs] def updateDataSeries(self, initialState=None): """Makes sure that a :class:`.DataSeries` instance has been created for every compatible overlay, and that property listeners are correctly registered, so the plot can be refreshed when needed. :arg initialState: If provided, must be a ``dict`` of ``{ overlay : bool }`` mappings, specifying the initial value of the :attr:`.DataSeries.enabled` property for newly created instances. If not provided, only the data series for the currently selected overlay (if it has been newly added) is initially enabled. """ # Default to showing the # currently selected overlay if initialState is None: if len(self.overlayList) > 0: initialState = {self.displayCtx.getSelectedOverlay() : True} else: initialState = {} # Make sure that a DataSeries # exists for every compatible overlay newOverlays = [] for ovl in self.overlayList: if ovl in self.__dataSeries: continue if isinstance(ovl, fsloverlay.ProxyImage): continue ds, refreshTargets, refreshProps = self.createDataSeries(ovl) display = self.displayCtx.getDisplay(ovl) if ds is None: # "Disable" overlays which don't have any data # to plot. We do this mostly so the overlay # appears greyed out in the OverlayListPanel. display.enabled = False continue # Display.enabled == DataSeries.enabled ds.bindProps('enabled', display) ds.enabled = initialState.get(ovl, False) log.debug('Created {} for overlay {} (enabled: {})'.format( type(ds).__name__, ovl, ds.enabled)) newOverlays.append(ovl) self.__dataSeries[ ovl] = ds self.__refreshProps[ovl] = (refreshTargets, refreshProps) # Make sure that property listeners are # registered all of these overlays for overlay in newOverlays: targets, propNames = self.__refreshProps.get(overlay, (None, None)) if targets is None: continue ds = self.__dataSeries[overlay] for propName in ds.redrawProperties(): ds.addListener(propName, self.__name, self.asyncDraw, overwrite=True) for target, propName in zip(targets, propNames): count = self.__refreshCounts.get((target, propName), 0) self.__refreshCounts[target, propName] = count + 1 if count == 0: log.debug('Adding listener on {}.{} for {} data ' 'series'.format(type(target).__name__, propName, overlay)) target.addListener(propName, self.__name, self.asyncDraw, overwrite=True)
[docs] @actions.toggleControlAction(overlaylistpanel.OverlayListPanel) def toggleOverlayList(self): """Shows/hides an :class:`.OverlayListPanel`. See :meth:`.ViewPanel.togglePanel`. """ # Tell the overlay list panel to disable # all overlays that aren't being plotted. # # This OverlayPlotPanel will always be # notified about a new overlay before # this OverlayListPanel, so a DataSeries # instance will always have been created # by the time the list panel calls this # filter function. def listFilter(overlay): return self.getDataSeries(overlay) is not None self.togglePanel(overlaylistpanel.OverlayListPanel, showVis=True, showSave=False, showGroup=False, propagateSelect=True, elistboxStyle=(elistbox.ELB_REVERSE | elistbox.ELB_TOOLTIP_DOWN | elistbox.ELB_NO_ADD | elistbox.ELB_NO_REMOVE | elistbox.ELB_NO_MOVE), location=wx.LEFT, filterFunc=listFilter)
[docs] @actions.toggleControlAction(plotlistpanel.PlotListPanel) def togglePlotList(self, floatPane=False): """Shows/hides a :class:`.PlotListPanel`. See :meth:`.ViewPanel.togglePanel`. """ self.togglePanel(plotlistpanel.PlotListPanel, self, location=wx.LEFT, floatPane=floatPane)
def __dataSeriesChanged(self, *a): """Called when the :attr:`dataSeries` list changes. Enables/disables the :meth:`removeDataSeries` action accordingly. """ self.removeDataSeries.enabled = len(self.dataSeries) > 0 def __overlayListChanged(self, *a, **kwa): """Called when the :class:`.OverlayList` changes. Makes sure that there are no :class:`.DataSeries` instances in the :attr:`.PlotPanel.dataSeries` list, or in the internal cache, which refer to overlays that no longer exist. :arg initialState: Must be passed as a keyword argument. If provided, passed through to the :meth:`updateDataSeries` method. """ initialState = kwa.get('initialState', None) for ds in list(self.dataSeries): if ds.overlay is not None and ds.overlay not in self.overlayList: self.dataSeries.remove(ds) ds.destroy() for overlay in list(self.__dataSeries.keys()): if overlay not in self.overlayList: self.clearDataSeries(overlay) for overlay in self.overlayList: display = self.displayCtx.getDisplay(overlay) # PlotPanels use the Display.enabled property # to toggle on/off overlay plots. We don't want # this to interfere with CanvasPanels, which # use Display.enabled to toggle on/off overlays. display.detachFromParent('enabled') self.updateDataSeries(initialState=initialState) self.asyncDraw()