matplotlib · ksunden · Apr 9, 2026 · Apr 7, 2026 · Apr 7, 2026
diff --git a/lib/matplotlib/backend_bases.py b/lib/matplotlib/backend_bases.py
@@ -540,19 +540,25 @@ def draw_text(self, gc, x, y, s, prop, angle, ismath=False, mtext=None):
         rendering backends, and indeed many builtin backends do not support
         this.  Rather, TeX rendering is provided by `~.RendererBase.draw_tex`.
         """
-        self._draw_text_as_path(gc, x, y, s, prop, angle, ismath)
+        self._draw_text_as_path(gc, x, y, s, prop, angle, ismath, mtext)
 
-    def _draw_text_as_path(self, gc, x, y, s, prop, angle, ismath):
+    def _draw_text_as_path(self, gc, x, y, s, prop, angle, ismath, mtext):
         """
         Draw the text by converting them to paths using `.TextToPath`.
 
         This private helper supports the same parameters as
         `~.RendererBase.draw_text`; setting *ismath* to "TeX" triggers TeX
         rendering.
         """
+        if mtext is not None:
+            features = mtext.get_fontfeatures()
+            language = mtext.get_language()
+        else:
+            features = language = None
         text2path = self._text2path
         fontsize = self.points_to_pixels(prop.get_size_in_points())
-        verts, codes = text2path.get_text_path(prop, s, ismath=ismath)
+        verts, codes = text2path.get_text_path(prop, s, ismath=ismath,
+                                               features=features, language=language)
         path = Path(verts, codes)
         if self.flipy():
             width, height = self.get_canvas_width_height()

diff --git a/lib/matplotlib/tests/baseline_images/test_text/draw_text_fallback.png b/lib/matplotlib/tests/baseline_images/test_text/draw_text_fallback.png
diff --git a/lib/matplotlib/tests/test_text.py b/lib/matplotlib/tests/test_text.py
@@ -115,26 +115,6 @@ def find_matplotlib_font(**kw):
     ax.set_yticks([])
 
 
-@image_comparison(['complex'], extensions=['png', 'pdf', 'svg', 'eps'], style='mpl20')
-def test_complex_shaping():
-    # Raqm is Arabic for writing; note that because Arabic is RTL, the characters here
-    # may seem to be in a different order than expected, but libraqm will order them
-    # correctly for us.
-    text = (
-        'Arabic: \N{Arabic Letter REH}\N{Arabic FATHA}\N{Arabic Letter QAF}'
-        '\N{Arabic SUKUN}\N{Arabic Letter MEEM}')
-    math_signs = '\N{N-ary Product}\N{N-ary Coproduct}\N{N-ary summation}\N{Integral}'
-    text = math_signs + text + math_signs
-    fig = plt.figure(figsize=(6, 2))
-    fig.text(0.5, 0.75, text, size=32, ha='center', va='center')
-    # Also check fallback behaviour:
-    # - English should use cmr10
-    # - Math signs should use DejaVu Sans Display (and thus be larger than the rest)
-    # - Arabic should use DejaVu Sans
-    fig.text(0.5, 0.25, text, size=32, ha='center', va='center',
-             family=['cmr10', 'DejaVu Sans Display', 'DejaVu Sans'])
-
-
 @image_comparison(['multiline'], style='mpl20')
 def test_multiline():
     plt.figure()
@@ -1244,10 +1224,31 @@ def test_text_tightbbox_outside_scale_domain():
     assert not np.isfinite(invalid_bbox.width)
 
 
-@image_comparison(['features'], remove_text=False, style='mpl20',
-                  extensions=['png', 'pdf', 'svg', 'eps'])
-def test_text_features():
-    fig = plt.figure(figsize=(5, 1.5))
+def _test_complex_shaping(fig):
+    # Raqm is Arabic for writing; note that because Arabic is RTL, the characters here
+    # may seem to be in a different order than expected, but libraqm will order them
+    # correctly for us.
+    text = (
+        'Arabic: \N{Arabic Letter REH}\N{Arabic FATHA}\N{Arabic Letter QAF}'
+        '\N{Arabic SUKUN}\N{Arabic Letter MEEM}')
+    math_signs = '\N{N-ary Product}\N{N-ary Coproduct}\N{N-ary summation}\N{Integral}'
+    text = math_signs + text + math_signs
+    fig.text(0.5, 0.75, text, size=32, ha='center', va='center')
+    # Also check fallback behaviour:
+    # - English should use cmr10
+    # - Math signs should use DejaVu Sans Display (and thus be larger than the rest)
+    # - Arabic should use DejaVu Sans
+    fig.text(0.5, 0.25, text, size=32, ha='center', va='center',
+             family=['cmr10', 'DejaVu Sans Display', 'DejaVu Sans'])
+
+
+@image_comparison(['complex'], extensions=['png', 'pdf', 'svg', 'eps'], style='mpl20')
+def test_complex_shaping():
+    fig = plt.figure(figsize=(6, 2))
+    _test_complex_shaping(fig)
+
+
+def _test_text_features(fig):
     t = fig.text(1, 0.7, 'Default: fi ffi fl st',
                  fontsize=32, horizontalalignment='right')
     assert t.get_fontfeatures() is None
@@ -1261,6 +1262,13 @@ def test_text_features():
     assert t.get_fontfeatures() == ('dlig', )
 
 
+@image_comparison(['features'], remove_text=False, style='mpl20',
+                  extensions=['png', 'pdf', 'svg', 'eps'])
+def test_text_features():
+    fig = plt.figure(figsize=(5, 1.5))
+    _test_text_features(fig)
+
+
 @pytest.mark.parametrize(
     'input, match',
     [
@@ -1275,11 +1283,7 @@ def test_text_language_invalid(input, match):
         Text(0, 0, 'foo', language=input)
 
 
-@image_comparison(['language'], remove_text=False, style='mpl20',
-                  extensions=['png', 'pdf', 'svg', 'eps'])
-def test_text_language():
-    fig = plt.figure(figsize=(5, 3))
-
+def _test_text_language(fig):
     t = fig.text(0, 0.8, 'Default', fontsize=32)
     assert t.get_language() is None
     t = fig.text(0, 0.55, 'Lang A', fontsize=32)
@@ -1317,3 +1321,22 @@ def test_text_language():
     t.set_language((('en', 0, 1), ('smn', 1, 2), ('en', 2, 3), ('smn', 3, 4)))
     assert t.get_language() == (
         ('en', 0, 1), ('smn', 1, 2), ('en', 2, 3), ('smn', 3, 4))
+
+
+@image_comparison(['language'], remove_text=False, style='mpl20',
+                  extensions=['png', 'pdf', 'svg', 'eps'])
+def test_text_language():
+    fig = plt.figure(figsize=(5, 3))
+    _test_text_language(fig)
+
+
+@image_comparison(['draw_text_fallback.png'], style='mpl20')
+def test_draw_text_as_path_fallback(monkeypatch):
+    # Delete RendererAgg.draw_text so that we use the RendererBase.draw_text fallback.
+    monkeypatch.delattr('matplotlib.backends.backend_agg.RendererAgg.draw_text')
+    heights = [2, 1.5, 3]
+    fig = plt.figure(figsize=(6, sum(heights)))
+    subfig = fig.subfigures(3, 1, height_ratios=heights)
+    _test_complex_shaping(subfig[0])
+    _test_text_features(subfig[1])
+    _test_text_language(subfig[2])