MNT: Make bar() error handling independent of numpy error message

as proposed in matplotlib#32431 (comment)

Author: timhoffm

lib/matplotlib/axes/_axes.py

@@ -2520,27 +2520,23 @@ def bar(self, x, height, width=0.8, bottom=None, *, align="center",
             height = self._convert_dx(height, y0, y, self.convert_yunits)
             if yerr is not None:
                 yerr = self._convert_dx(yerr, y0, y, self.convert_yunits)
+
+        # broadcast all args to make them iterable
+        x = np.atleast_1d(x)
+        args = (x, height, width, y, linewidth, hatch)
+        arg_names = ('x', 'height', 'width', 'y', 'linewidth', 'hatch')
         try:
-            x, height, width, y, linewidth, hatch = np.broadcast_arrays(
-                # Make args iterable too.
-                np.atleast_1d(x), height, width, y, linewidth, hatch
-            )
+            x, height, width, y, linewidth, hatch = np.broadcast_arrays(*args)
         except ValueError as e:
-            arg_map = {
-                "arg 0": "'x'",
-                "arg 1": "'height'",
-                "arg 2": "'width'",
-                "arg 3": "'y'",
-                "arg 4": "'linewidth'",
-                "arg 5": "'hatch'"
-            }
-            error_message = str(e)
-            for arg, name in arg_map.items():
-                error_message = error_message.replace(arg, name)
-            if error_message != str(e):
-                raise ValueError(error_message) from e
-            else:
-                raise
+            non_trivial_inputs = ", ".join(
+                f"{name} {np.shape(arg)}" for arg, name in zip(args, arg_names)
+                # only collect args that could contribute to a broadcast error
+                if any(dim != 1 for dim in np.shape(arg))
+            )
+            raise ValueError(
+                f"bar() failed on broadcasting inputs with the following shapes: "
+                f"{non_trivial_inputs}"
+            ) from e
 
         # Now that units have been converted, set the tick locations.
         if orientation == 'vertical':

lib/matplotlib/tests/test_axes.py

@@ -10411,7 +10411,8 @@ def test_bar_shape_mismatch():
     x = ["foo", "bar"]
     height = [1, 2, 3]
     error_message = (
-        r"Mismatch is between 'x' with shape \(2,\) and 'height' with shape \(3,\)"
+        r"bar\(\) failed on broadcasting inputs with the following shapes: "
+        r"x \(2,\), height \(3,\)"
     )
     with pytest.raises(ValueError, match=error_message):
         plt.bar(x, height)