Merge pull request matplotlib#23090 from anntzer/nfm

Derive new_figure_manager from FigureCanvas.new_manager.

Author: tacaswell

lib/matplotlib/pyplot.py

@@ -266,15 +266,9 @@ def switch_backend(newbackend):
             rcParamsOrig["backend"] = "agg"
             return
 
-    # Backends are implemented as modules, but "inherit" default method
-    # implementations from backend_bases._Backend.  This is achieved by
-    # creating a "class" that inherits from backend_bases._Backend and whose
-    # body is filled with the module's globals.
-
-    backend_name = cbook._backend_module_name(newbackend)
-
-    class backend_mod(matplotlib.backend_bases._Backend):
-        locals().update(vars(importlib.import_module(backend_name)))
+    backend_mod = importlib.import_module(
+        cbook._backend_module_name(newbackend))
+    canvas_class = backend_mod.FigureCanvas
 
     required_framework = _get_required_interactive_framework(backend_mod)
     if required_framework is not None:
@@ -286,6 +280,36 @@ class backend_mod(matplotlib.backend_bases._Backend):
                 "framework, as {!r} is currently running".format(
                     newbackend, required_framework, current_framework))
 
+    # Load the new_figure_manager(), draw_if_interactive(), and show()
+    # functions from the backend.
+
+    # Classically, backends can directly export these functions.  This should
+    # keep working for backcompat.
+    new_figure_manager = getattr(backend_mod, "new_figure_manager", None)
+    # draw_if_interactive = getattr(backend_mod, "draw_if_interactive", None)
+    # show = getattr(backend_mod, "show", None)
+    # In that classical approach, backends are implemented as modules, but
+    # "inherit" default method implementations from backend_bases._Backend.
+    # This is achieved by creating a "class" that inherits from
+    # backend_bases._Backend and whose body is filled with the module globals.
+    class backend_mod(matplotlib.backend_bases._Backend):
+        locals().update(vars(backend_mod))
+
+    # However, the newer approach for defining new_figure_manager (and, in
+    # the future, draw_if_interactive and show) is to derive them from canvas
+    # methods.  In that case, also update backend_mod accordingly.
+    if new_figure_manager is None:
+        def new_figure_manager_given_figure(num, figure):
+            return canvas_class.new_manager(figure, num)
+
+        def new_figure_manager(num, *args, FigureClass=Figure, **kwargs):
+            fig = FigureClass(*args, **kwargs)
+            return new_figure_manager_given_figure(num, fig)
+
+        backend_mod.new_figure_manager_given_figure = \
+            new_figure_manager_given_figure
+        backend_mod.new_figure_manager = new_figure_manager
+
     _log.debug("Loaded backend %s version %s.",
                newbackend, backend_mod.backend_version)
 

lib/matplotlib/testing/conftest.py

@@ -44,13 +44,13 @@ def mpl_test_settings(request):
 
         backend = None
         backend_marker = request.node.get_closest_marker('backend')
+        prev_backend = matplotlib.get_backend()
         if backend_marker is not None:
             assert len(backend_marker.args) == 1, \
                 "Marker 'backend' must specify 1 backend."
             backend, = backend_marker.args
             skip_on_importerror = backend_marker.kwargs.get(
                 'skip_on_importerror', False)
-            prev_backend = matplotlib.get_backend()
 
             # special case Qt backend importing to avoid conflicts
             if backend.lower().startswith('qt5'):
@@ -87,8 +87,7 @@ def mpl_test_settings(request):
         try:
             yield
         finally:
-            if backend is not None:
-                plt.switch_backend(prev_backend)
+            matplotlib.use(prev_backend)
 
 
 @pytest.fixture

lib/matplotlib/tests/test_backend_template.py

@@ -0,0 +1,23 @@
+"""
+Backend-loading machinery tests, using variations on the template backend.
+"""
+
+import sys
+from types import SimpleNamespace
+
+import matplotlib as mpl
+from matplotlib import pyplot as plt
+from matplotlib.backends import backend_template
+
+
+def test_load_template():
+    mpl.use("template")
+    assert type(plt.figure().canvas) == backend_template.FigureCanvasTemplate
+
+
+def test_new_manager(monkeypatch):
+    mpl_test_backend = SimpleNamespace(**vars(backend_template))
+    del mpl_test_backend.new_figure_manager
+    monkeypatch.setitem(sys.modules, "mpl_test_backend", mpl_test_backend)
+    mpl.use("module://mpl_test_backend")
+    assert type(plt.figure().canvas) == backend_template.FigureCanvasTemplate