如何更改python中“保存图形”的默认路径？

Question

I have a python code to create a figure. After showing it with plt.show() , I want to save the figure.
To avoid messing up the aspect ratio, resolution, etc, I do not want to use the savefig -command in the code. Instead, I want to use the "save the figure" button from the figure window.
However, by default, it prompts my home folder as location for the save. I would like the save to automatically be in the directory where the code was executed.
How/where can I change this window default path for saving to the current folder (or somewhere else)?

I tried this command from Change directory to the directory of a Python script at the beginning but it did not help, even though gives the filename correctly:

os.chdir(os.path.dirname(__file__))

Answer 1

It looks like you may be able to set this by changing the defaults file matplotlibrc , check out the guidance under http://matplotlib.org/users/customizing.html where the important lines are under the savefig parameters:

# the default savefig params can be different from the display params

...

savefig.directory   : ~        # default directory in savefig dialog box, 
                               # leave empty to always use current working directory

It seems this was introduced in matplotlib 1.3. I guess you could set this using,

matplotlib as mpl
mpl.rcParams["savefig.directory"] = os.chdir(os.path.dirname(__file__))

at the top of a script or by changing the matplotlibrc file. For the dialog to default to cwd instead of script location (thanks to jjcf89 for this)

matplotlib as mpl 
mpl.rcParams["savefig.directory"] = ""

Answer 2

For a non-code solution, you can create a file called matplotlibrc with the following contents:

savefig.directory:

This sets the value of savefig.directory to the empty string, causing it default to the current working directory.

You can put this matplotlib file in the current working directory, or in a location specified in the documentation . For example, on Linux it should be in .config/matplotlib/ (don't be worried if there's no matplotlibrc file there already, there wasn't one on my machine either).