Matplotlib can't suppress figure window

Magic_Matt_Man picture Magic_Matt_Man · Nov 17, 2014 · Viewed 11.9k times · Source

I'm having trouble with matplotlib insisting on displaying a figure wnidow even when I haven't called show().

The function in question is:

def make_plot(df):
    fig, axes = plt.subplots(3, 1, figsize=(10, 6), sharex=True)
    plt.subplots_adjust(hspace=0.2)

    axes[0].plot(df["Date_Time"], df["T1"], df["Date_Time"], df["T2"])
    axes[0].set_ylabel("Temperature (C)")
    axes[0].legend(["T1", "T2"], bbox_to_anchor=(1.12, 1.1))
    axes[1].semilogy(df["Date_Time"], df["IGP"], df["Date_Time"], df["IPP"])
    axes[1].legend(["IGP", "IPP"], bbox_to_anchor=(1.12, 1.1))
    axes[1].set_ylabel("Pressure (mBar)")
    axes[2].plot(df["Date_Time"], df["Voltage"], "k")
    axes[2].set_ylabel("Voltage (V)")
    current_axes = axes[2].twinx()
    current_axes.plot(df["Date_Time"], df["Current"], "r")
    current_axes.set_ylabel("Current (mA)")
    axes[2].legend(["V"], bbox_to_anchor=(1.15, 1.1))
    current_axes.legend(["I"], bbox_to_anchor=(1.14, 0.9))

    plt.savefig("static/data.png")

where df is a dataframe created using pandas. This is supposed to be in the background of a web server, so all I want is for this function to drop the file in the directory specified. However, when it executes it does this, and then pulls up a figure window and gets stuck in a loop, preventing me from reloading the page. Am I missing something obvious?

EDIT: Forgot to add, I am running python 2.7 on Windows 7, 64 bit.

Answer

user707650 picture user707650 · Nov 17, 2014

Step 1

Check whether you're running in interactive mode. The default is non-interactive, but you may never know:

>>> import matplotlib as mpl
>>> mpl.is_interactive()
False

You can set the mode explicitly to non-interactive by using

>>> from matplotlib import pyplot as plt
>>> plt.ioff()

Since the default is non-interactive, this is probably not the problem.

Step 2

Make sure your backend is a non-gui backend. It's the difference between using Agg versus TkAgg, WXAgg, GTKAgg etc, the latter being gui backends, while Agg is a non-gui backend.

You can set the backend in a number of ways:

  • in your matplotlib configuration file; find the line starting with backend:

    backend: Agg
    
  • at the top of your program with the global matplotlib function use:

    matplotlib.use('Agg')
    
  • import the canvas directly from the correct backend; this is most useful in non-pyplot "mode" (OO-style), which is what I often use, and for a webserver style of use, that may in the end prove best (since this is a tad different than above, here's a full-blown short example):

    import numpy as np
    from matplotlib.figure import Figure
    from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
    figure = Figure()
    canvas = FigureCanvas(figure)
    axes = figure.add_subplot(1, 1, 1)
    axes.plot(x, np.sin(x), 'k-')
    canvas.print_figure('sine.png')