Transform aesthetic parameters

doc/changelog.qmd

@@ -13,6 +13,16 @@ title: Changelog
 - Fixed [](:class:`~plotnine.geom_text`) with `adjust_text` for some
   cases where the text are placed outside the panels. {{< issue 899 >}}
 
+- The default aesthetics and aesthetic parameters of geoms (and stats) are
+  now transformed. So you can now set date values as aesthetic parameters.
+  e.g.
+
+  ```python
+  geom_point(y=datetime(2024, 12, 11))
+  ```
+
+  if the `y` aesthetic is mapped to datetime column in another layer.
+
 ## v0.14.3
 (2024-11-26)
 [![](https://zenodo.org/badge/DOI/10.5281/zenodo.14224336.svg)](https://doi.org/10.5281/zenodo.14224336)

plotnine/geoms/geom.py

@@ -217,7 +217,7 @@ def use_defaults(
         data :
             Data used for drawing the geom.
         aes_modifiers :
-            Aesthetics
+            Aesthetics to evaluate
 
         Returns
         -------

plotnine/ggplot.py

@@ -406,7 +406,7 @@ def _build(self):
         layout.setup_panel_params(self.coordinates)
 
         # fill in the defaults
-        layers.use_defaults_after_scale()
+        layers.use_defaults_after_scale(scales)
 
         # Allow stats to modify the layer data
         layers.finish_statistics()

plotnine/layer.py

@@ -370,6 +370,7 @@ def use_defaults(
         self,
         data: pd.DataFrame,
         aes_modifiers: dict[str, Any],
+        scales: Scales | None = None,
     ) -> pd.DataFrame:
         """
         Prepare/modify data for plotting
@@ -382,7 +383,17 @@ def use_defaults(
             Expression to evaluate and replace aesthetics in
             the data.
         """
-        return self.geom.use_defaults(data, aes_modifiers)
+        old_columns = data.columns
+        data = self.geom.use_defaults(data, aes_modifiers)
+        if scales is not None:
+            # The default aesthetics and the aesthetic parameters are
+            # specified in userspace. When we add them we have to
+            # transform them.
+            new_columns = data.columns.difference(old_columns)
+            _data = scales.transform_df(self.data[new_columns])
+            for col in new_columns:
+                data[col] = _data[col]
+        return data
 
     def finish_statistics(self):
         """
@@ -468,9 +479,9 @@ def compute_position(self, layout: Layout):
         for l in self:
             l.compute_position(layout)
 
-    def use_defaults_after_scale(self):
+    def use_defaults_after_scale(self, scales: Scales):
         for l in self:
-            l.use_defaults(l.data, l.mapping._scaled)
+            l.data = l.use_defaults(l.data, l.mapping._scaled, scales)
 
     def transform(self, scales: Scales):
         for l in self:

tests/test_scale_internals.py

@@ -53,6 +53,7 @@
     scale_x_discrete,
     scale_x_log10,
     scale_y_continuous,
+    scale_y_log10,
 )
 from plotnine.scales.scales import make_scale
 
@@ -876,3 +877,33 @@ def test_discrete_scale_for_empty_layer():
     )
 
     p.draw_test()
+
+
+def test_transform_aes_defaults_and_params():
+    data = pd.DataFrame({"x": range(5), "y": 10 ** np.arange(5)})
+    # Another set of points that should be inline with the first set
+    # both with and without the scale_y_log10.
+    data2 = pd.DataFrame({"x": [3.2, 3.4, 3.6, 3.8]})
+    yparam = list(10 ** data2["x"])
+
+    p = (
+        ggplot(data, aes("x", "y"))
+        + geom_point()
+        + geom_point(data=data2, y=yparam, color="red")
+        + scale_y_log10()
+    )
+    assert p == "transform_aes_defaults_and_params"
+
+
+def test_transform_datetime_aes_param():
+    data = pd.DataFrame(
+        {"x": range(5), "y": [datetime(2024, i, 1) for i in range(1, 6)]}
+    )
+    yparam: list[datetime] = list(data["y"] + timedelta(days=30))
+
+    p = (
+        ggplot(data, aes("x", "y"))
+        + geom_point()
+        + geom_point(y=yparam, color="red")
+    )
+    assert p == "transform_datetime_aes_param"