Adding an item in matplotlib´s toolbar

Based on this related SO answer, I wanted a way that doesn't involve extending any class and is much less verbose (pretty much 3 lines). Taking a look to this sources and using fig.canvas.toolbar I made this snippet work:

It plots an image and adds two buttons to the toolbar: previous and next, which can be bound to any parameter-less function. This can be very convenient, for instance, for scrolling through the images on a folder and drawing some points or lines on them using scatter, as explained in this other SO post. Let me know how it works for you!

fig, ax = plt.subplots()
fig.canvas.manager.toolbar._Button("PREVIOUS", "back_large", <ACTION_PREV>)
fig.canvas.manager.toolbar._Button("NEXT", "forward_large", <ACTION_NEXT>)
im = plt.imread("hello.png")
implot = ax.imshow(im)
fig.show()

• Note the bigger arrows that were added to the toolbar.

• The drawback of this method is that it uses the protected _Button method, whose interface may change in further versions, but this way seems fine to me for some casual scripting.

• Also note that the method signature (see the sources) requires a path to an image that (in my case) has to be found at /usr/local/lib/python2.7/dist-packages/matplotlib/mpl-data/images. Again, there is probably a way to overcome this but probably more verbose than the inheriting alternative.

Cheers,
Andres

BONUS:

class ImageViewer(object):
    """Given a path to a directory, this class opens a matplotib window with two
       custom buttons that allow scrolling through the images in the directory.
       Usage example: iv = ImageViewer("/home/pau/Images")
    """
    def __init__(self, seq_path, img_extension=".png"):
        self.ids = [join(seq_path, splitext(img)[0]) for img in listdir(seq_path)
                    if img.endswith(img_extension)]
        self.mod = len(self.ids)
        self.idx = 0
        self.fig, self.ax = plt.subplots()
        self.fig.canvas.manager.toolbar._Button("PREVIOUS", "back_large", self.prev)
        self.fig.canvas.manager.toolbar._Button("NEXT", "forward_large", self.next)
        self._plot(0)
    def _plot(self, idx):
        im = plt.imread(self.ids[idx]+".png")
        implot = self.ax.imshow(im, cmap='Greys_r',  interpolation='nearest')
        self.fig.show()
    def next(self):
        self.idx = (self.idx+1)%self.mod
        self._plot(self.idx)
    def prev(self):
        self.idx = (self.idx-1)%self.mod
        self._plot(self.idx)

I have solved the problem. I´ ve found this code:

Matplotlib/Tkinter - customizing toolbar tooltips (broken link)

So, I created a subclass and added as it says in the link. This is the code:

class MyToolbar(NavigationToolbar2):
  def __init__(self, figure_canvas, parent= None):
    self.toolitems = (('Home', 'Lorem ipsum dolor sit amet', 'home', 'home'),
        ('Back', 'consectetuer adipiscing elit', 'back', 'back'),
        ('Forward', 'sed diam nonummy nibh euismod', 'forward', 'forward'),
        (None, None, None, None),
        ('Pan', 'tincidunt ut laoreet', 'move', 'pan'),
        ('Zoom', 'dolore magna aliquam', 'zoom_to_rect', 'zoom'),
        (None, None, None, None),
        ('Subplots', 'putamus parum claram', 'subplots', 'configure_subplots'),
        ('Save', 'sollemnes in futurum', 'filesave', 'save_figure'),
        ('Port', 'Select', "select", 'select_tool'),
        )

    NavigationToolbar2.__init__(self, figure_canvas, parent= None)

  def select_tool(self):
    print "You clicked the selection tool"

And the, you can add this toolbar by writing:

self.navigation_toolbar = MyToolbar(self.figure_canvas, self)            
self.navigation_toolbar.update()

If you want to only let your own button, you must erase all the others items from self.toolitems. For example:

self.toolitems = (
    ('Port', 'Select', "select", 'select_tool'),
    )

With this, you will only see your own button in the NavigationToolbar

Hope this helps.

As noted, if you don't mind extending the class you can add buttons to the toolbar by taking the list of toolitems and modifying it accordingly. To get a custom icon for the new toolbar item for that particular approach, use something like this for a NewButton:

class MyNavigationBar(NavigationToolbar2Tk):

    def __init__(self,canvas,frame,pack_toolbar=False):
        self.toolitems = (
            ('Home', 'Reset original view', 'home', 'home'),
            ('Back', 'Back to previous view', 'back', 'back'),
            ('Forward', 'Forward to next view', 'forward', 'forward'),
            (None, None, None, None),
            ('Pan',
            'Left button pans, Right button zooms\n'
            'x/y fixes axis, CTRL fixes aspect',
            'move', 'pan'),
            ('NewButton', 'Adjust window and level','filesave','windowlevel'),
            (None, None, None, None),
            ('Save', 'Save the figure', 'filesave', 'save_figure'),
        )
        super().__init__(canvas,frame,pack_toolbar=pack_toolbar)
        path = os.path.join('/path/to/icondir','tool_icon.png')
        self._buttons['NewButton']._image_file = path
        NavigationToolbar2Tk._set_image_for_button(self, self._buttons['NewButton'])

   def windowlevel(self,*args):
        do something...

note the use of an existing tk navigation bar icon filesave in the list above for NewButton, it is just a placeholder until it gets replaced by the custom icon.