dvg-pyqtgraph-threadsafe 3.4.0

PyQtGraph library providing thread-safe plot curves with underlying (ring) buffers.

DvG_PyQtGraph_ThreadSafe

PyQtGraph library providing thread-safe plot curves with underlying (ring) buffers.

Supports PyQt5, PyQt6, PySide2 and PySide6.

• Github: https://github.com/Dennis-van-Gils/python-dvg-pyqtgraph-threadsafe

• PyPI: https://pypi.org/project/dvg-pyqtgraph-threadsafe

Installation:

pip install dvg-pyqtgraph-threadsafe

You must also ensure a Qt library is installed in your Python environment as this library will not install one for you. Pick one (personal recommendation PySide6):

pip install pyqt5
pip install pyqt6
pip install pyside2
pip install pyside6

Futhermore, you might want to enable OpenGL hardware accelerated plotting by installing PyOpenGL:

pip install pyopengl

Useful links

• Demos can be found in the demos folder.

• Benchmark tests and results can be found in the benchmark folder.

• Important notes on the specific use-case where another thread is actively requesting data at a high rate from an external device (e.g. an Arduino) to be pushed into a ThreadSafeCurve can be found here.

Overview

Classes HistoryChartCurve, BufferedPlotCurve & PlotCurve wrap around a pyqtgraph.PlotDataItem instance, called a curve for convenience. Data can be safely appended or set from out of any thread.

The (x, y)-curve data is buffered internally to the class, relying on either a circular/ring buffer or a regular array buffer:

• HistoryChartCurve

  Ring buffer. The plotted x-data will be shifted such that the right-side is always set to 0. I.e., when x denotes time, the data is plotted backwards in time, hence the name history chart. The most recent data is on the right-side of the ring buffer.

• BufferedPlotCurve

  Ring buffer. Data will be plotted as is. Can also act as a Lissajous figure.

• PlotCurve

  Regular array buffer. Data will be plotted as is.

Usage

import sys

from qtpy import QtWidgets
import pyqtgraph as pg

from dvg_pyqtgraph_threadsafe import HistoryChartCurve

class MainWindow(QtWidgets.QWidget):
    def __init__(self, parent=None, **kwargs):
        super().__init__(parent, **kwargs)

        self.gw = pg.GraphicsLayoutWidget()
        self.plot_1 = self.gw.addPlot()

        # Create a HistoryChartCurve and have it wrap around a new
        # PlotDataItem as set by argument `linked_curve`.
        self.tscurve_1 = HistoryChartCurve(
            capacity=5,
            linked_curve=self.plot_1.plot(pen=pg.mkPen('r')),
        )

        grid = QtWidgets.QGridLayout(self)
        grid.addWidget(self.gw)

app = QtWidgets.QApplication(sys.argv)
window = MainWindow()

# The following line could have been executed from inside of another thread:
window.tscurve_1.extendData([1, 2, 3, 4, 5], [10, 20, 30, 20, 10])

# Draw the curve from out of the main thread
window.tscurve_1.update()

window.show()
sys.exit(app.exec())

Benchmark

If you intend to use this library to plot 100.000s of points at a high frame rate on your integrated GPU instead of on a dedicated (performance) GPU, you might run into performance issues. Even when OpenGL hardware acceleration is enabled.

I recommend you run the performance benchmark found in the benchmark folder. It can test for different PyQtGraph versions and Qt libraries on your GPU of choice. I have found that the older pyqtgraph 0.11 library sometimes results in a better performance than pyqtgraph 0.12 on integrated GPUs. See benchmark results.

API

Class ThreadSafeCurve

ThreadSafeCurve(
    capacity: int | None,
    linked_curve: pyqtgraph.PlotDataItem,
    shift_right_x_to_zero: bool = False,
    use_ringbuffer=None,
)

Note

Provides the base class for a thread-safe plot curve to which (x, y)-data can be safely appended or set from out of any thread. It will wrap around the passed argument linked_curve of type pyqtgraph.PlotDataItem and will manage the (x, y)-data buffers underlying the curve.

Intended multi-threaded operation: One or more threads push new data into the ThreadSafeCurve-buffers. Another thread performs the GUI refresh by calling update() which will redraw the curve according to the current buffer contents.

Args:

capacity (int, optional):

When an integer is supplied it defines the maximum number op points each of the x-data and y-data buffers can store. The x-data buffer and the y-data buffer are each a ring buffer. New readings are placed at the end (right-side) of the buffer, pushing out the oldest readings when the buffer has reached its maximum capacity (FIFO). Use methods appendData() and extendData() to push in new data.

When None is supplied the x-data and y-data buffers are each a regular array buffer of undefined length. Use method setData() to set the data.

linked_curve (pyqtgraph.PlotDataItem):

Instance of pyqtgraph.PlotDataItem to plot the buffered data out into.

shift_right_x_to_zero (bool, optional):

When plotting, should the x-data be shifted such that the right-side is always set to 0? Useful for history charts.

Default: False

use_ringbuffer (bool, deprecated):

Deprecated since v3.1.0. Defined for backwards compatibility. Simply supply a value for capacity to enable use of a ring buffer.

Attributes:

x_axis_divisor (float):

The x-data in the buffer will be divided by this factor when the plot curve is drawn. Useful to, e.g., transform the x-axis units from milliseconds to seconds or minutes.

Default: 1

y_axis_divisor (float):

Same functionality as x_axis_divisor.

Default: 1

Methods:

• appendData(x, y)

  Append a single (x, y)-data point to the ring buffer.

• extendData(x_list, y_list)

  Extend the ring buffer with a list of (x, y)-data points.

• setData(x_list, y_list)

  Set the (x, y)-data of the regular array buffer.

• update(create_snapshot: bool = True)

  Update the data behind the curve by creating a snapshot of the current contents of the buffer, and redraw the curve on screen.

  You can suppress updating the data behind the curve by setting parameter create_snapshot to False. The curve will then only be redrawn based on the old data. This is useful when the plot is paused.

• clear()

  Clear the contents of the curve and redraw.

• name()

  Get the name of the curve.

• isVisible() -> bool

  Is the curve visible?

• setVisible(state: bool = True)

  Set the visibility of the curve.

• setDownsampling(*args, **kwargs)

  All arguments will be passed onto method pyqtgraph.PlotDataItem.setDownsampling() of the underlying curve.

Properties:

• size -> Tuple[int, int]:

  Number of elements currently contained in the underlying (x, y)- buffers of the curve. Note that this is not necessarily the number of elements of the currently drawn curve. Instead, it reflects the current sizes of the data buffers behind it that will be drawn onto screen by the next call to update().

Class HistoryChartCurve

HistoryChartCurve(
    capacity: int,
    linked_curve: pyqtgraph.PlotDataItem,
)

Note

Inherits from: ThreadSafeCurve

Provides a thread-safe curve with underlying ring buffers for the (x, y)-data. New readings are placed at the end (right-side) of the buffer, pushing out the oldest readings when the buffer has reached its maximum capacity (FIFO). Use methods appendData() and extendData() to push in new data.

The plotted x-data will be shifted such that the right-side is always set to 0. I.e., when x denotes time, the data is plotted backwards in time, hence the name history chart.

See class ThreadSafeCurve for more details.

Class BufferedPlotCurve

BufferedPlotCurve(
    capacity: int,
    linked_curve: pyqtgraph.PlotDataItem,
)

Note

Inherits from: ThreadSafeCurve

Provides a thread-safe curve with underlying ring buffers for the (x, y)-data. New readings are placed at the end (right-side) of the buffer, pushing out the oldest readings when the buffer has reached its maximum capacity (FIFO). Use methods appendData() and extendData() to push in new data.

See class ThreadSafeCurve for more details.

Class PlotCurve

PlotCurve(
    capacity: int,
    linked_curve: pyqtgraph.PlotDataItem,
)

Note

Inherits from: ThreadSafeCurve

Provides a thread-safe curve with underlying regular array buffers for the (x, y)-data. Use method setData() to set the data.

See class ThreadSafeCurve for more details.

API Extras

Class LegendSelect

LegendSelect(
    linked_curves: Sequence[pyqtgraph.PlotDataItem | ThreadSafeCurve],
    hide_toggle_button: bool = False,
    box_bg_color: QtGui.QColor = QtGui.QColor(0, 0, 0),
    box_width: int = 40,
    box_height: int = 23,
    parent=None,
)

Note

Inherits from: PyQt5.QtCore.QObject

Creates and manages a legend of all passed curves with checkboxes to show or hide each curve. The legend ends with a push button to show or hide all curves in one go. The full set of GUI elements is contained in attribute grid of type PyQt5.QtWidget.QGridLayout to be added to your GUI.

Example grid:

□ Curve 1  [  /  ]
□ Curve 2  [  /  ]
□ Curve 3  [  /  ]
[ Show / Hide all]

The initial visibility, name and pen of each curve will be retrieved from the members within the passed curves, i.e.:

• curve.isVisible()

• curve.name()

• curve.opts["pen"]

Args:

linked_curves (Sequence[pyqtgraph.PlotDataItem | ThreadSafeCurve]):

Sequence of pyqtgraph.PlotDataItem or ThreadSafeCurve instances to be controlled by the legend.

hide_toggle_button (bool, optional):

Default: False

box_bg_color (QtGui.QColor, optional):

Background color of the legend boxes.

Default: QtGui.QColor(0, 0, 0)

box_width (int, optional):

Default: 40

box_height (int, optional):

Default: 23

Attributes:

chkbs (List[PyQt5.QtWidgets.QCheckbox]):

List of checkboxes to control the visiblity of each curve.

painted_boxes (List[PyQt5.QtWidgets.QWidget]):

List of painted boxes illustrating the pen of each curve.

qpbt_toggle (PyQt5.QtWidgets.QPushButton):

Push button instance that toggles showing/hiding all curves in one go.

grid (PyQt5.QtWidgets.QGridLayout):

The full set of GUI elements combined into a grid to be added to your GUI.

Class PlotManager

PlotManager(
    parent=None,
)

Note

Inherits from: PyQt5.QtCore.QObject

Creates and manages a collection of pushbuttons with predefined actions operating on the linked plots and curves. The full set of pushbuttons is contained in attribute grid of type PyQt5.QtWidget.QGridLayout to be added to your GUI.

Example grid:

[   Full range  ]
[auto x] [auto y]
[      0:30     ]
[      1:00     ]
[      3:00     ]
[     10:00     ]

[     Clear     ]

The grid starts empty and is build up by calling the following methods:

• add_autorange_buttons(): Adds the [Full range], [auto x] and [auto y] buttons.

• add_preset_buttons(): Adds presets on the x-axis range to zoom to.

• add_clear_button(): Adds the ‘Clear’ button.

Args:

parent (PyQt5.QtWidgets.QWidget):

Needs to be set to the parent QWidget for the QMessageBox as fired by button Clear to appear centered and modal to.

Attributes:

grid (PyQt5.QtWidgets.QGridLayout):

The full set of pushbuttons combined into a grid to be added to your GUI.

Changelog

3.4.0 (2024-06-24)

Code quality improvements:

• Using qtpy library instead of my own Qt5/6 mechanism

• Using f-strings

• Extended API and docstrings, like on PlotManager.

• Arguments linked_curves and linked_plots are hinted as Sequence instead of as List

Added support:

• Support for Numpy 2.0

3.3.0 (2023-02-27)

• Raise ImportError instead of general Exception

3.2.6 (2022-10-13)

• Added link to notes on use-case: DAQ

3.2.5 (2022-10-13)

• Minor edit: Using explicit arguments x and y to set the curve data and set the flag skipFiniteCheck=True. Both save (marginal) CPU time.

3.2.4 (2022-10-12)

• Bug fix: Snapshot creation checks if internal buffer is of type RingBuffer and casts to numpy.ndarray. This resolves an intermittent error where the isfinite boolean return array was operating as mask on the RingBuffer, which fails. The boolean return array now always operates on numpy.ndarray.

• Fix for external bug in paintGL of pyqtgraphgraphicsItemsPlotCurveItem.py: Added explicit check to ensure that the curve is only drawn when it has more than 1 point.

• Added benchmark running python=3.10 and pyqtgraph==0.13.1

3.2.3 (2022-10-11)

• Fixed bug when using PyQt6: QMessageBox.Yes –> QMessageBox.StandardButton.Yes

3.2.2 (2022-09-18)

• Updated discussion on benchmarks

• The batch file for auto-creation of the conda environments found in the benchmark folder now uses the conda-forge channel.

3.2.1 (2022-09-17)

• Neater mechanism to support both PyQt and PySide

• Added benchmarks

• Improved demos

3.2.0 (2022-09-13)

• Added support for PyQt5, PyQt6, PySide2 and PySide6 as suggested via a pull request by Mathijs van Gorcum (https://github.com/mvgorcum).

3.1.0 (2021-05-10)

• Removed redundant argument capacity from init PlotCurve(). It is not using a ringbuffer and, hence, does not need a capacity.

3.0.1 (2020-08-07)

Bug-fixes:

• The use of typing.TypedDict broke support under Python 3.6 and 3.7. Fixed by conditional import typing_extensions.

• Curve plotting was broken when setClipToView(True) and the curve data extended past the viewbox limits, when not using OpenGL. The cause was my incorrect calculation of connect. Fixed by commenting out connect again. Curves will now show continuously (linear interpolation) whenever a NaN is encountered, instead of as correctly fragmented. That’s acceptable.

3.0.0 (2020-08-07)

• Renamed parameter LegendSelect.curves to LegendSelect.linked_curves

• Changed base of class LegendSelect() from QWidget to QObject

• Added class PlotManager()

2.0.1 (2020-08-03)

• Workaround: PyQt5 >= 5.12.3 causes a bug in PyQtGraph where a curve won’t render if it contains NaNs (but only in the case when OpenGL is disabled). The curve will now be displayed correctly, i.e., fragmented whenever a NaN is encountered. When OpenGL is enabled, linear interpolation will occur at the gaps as per pyqtgraph.plotCurveItem.paintGL().

2.0.0 (2020-08-02)

• Method names are now conform the PyQtGraph naming style. I.e. setData() vs. set_data(), etc.

• The default values of PyQtGraph are no longer being overwritten.

• Added class LegendSelect

1.0.0 (2020-07-30)

• First release on PyPI