fix(datashader): render_shapes in a non-global coordinate system (#782)

src/spatialdata_plot/pl/render.py

@@ -895,11 +895,17 @@ def _draw_centroids(xy: np.ndarray, radius: float | None = None) -> None:
         transformed_geometry = shapes["geometry"].transform(
             lambda x: (np.hstack([x, np.ones((x.shape[0], 1))]) @ tm.T)[:, :2]
         )
+        transformed_geodataframe = gpd.GeoDataFrame(
+            data=shapes.drop("geometry", axis=1),
+            geometry=transformed_geometry,
+        )
+        # Geometry is already in the target CS, so tag the wrapped element with an identity transform
+        # there. Depending on the pandas/geopandas version the constructor above may carry over the
+        # source element's transform attrs; drop them first so parse does not see conflicting ones.
+        transformed_geodataframe.attrs.pop("transform", None)
         transformed_element = ShapesModel.parse(
-            gpd.GeoDataFrame(
-                data=shapes.drop("geometry", axis=1),
-                geometry=transformed_geometry,
-            )
+            transformed_geodataframe,
+            transformations={coordinate_system: Identity()},
         )
 
         if len(transformed_element) == 0:
@@ -908,7 +914,7 @@ def _draw_centroids(xy: np.ndarray, radius: float | None = None) -> None:
             return
 
         plot_width, plot_height, x_ext, y_ext, factor = _get_extent_and_range_for_datashader_canvas(
-            transformed_element, "global", fig_params
+            transformed_element, coordinate_system, fig_params
         )
 
         cvs = ds.Canvas(plot_width=plot_width, plot_height=plot_height, x_range=x_ext, y_range=y_ext)

tests/pl/test_render_shapes.py

@@ -1331,10 +1331,10 @@ def test_plot_can_handle_non_numeric_radius_values(sdata_blobs: SpatialData):
 def test_groups_filtering_preserves_transformation(sdata_blobs: SpatialData):
     """Regression test for #420: groups filtering must not strip coordinate-system metadata.
 
-    Simulates the exact sequence that ``_render_shapes`` performs —
+    Simulates the sequence that ``_render_shapes`` performs —
     filter_by_coordinate_system -> groups boolean-index -> reset_index ->
-    re-assign to sdata_filt -> GeoDataFrame re-wrap — then asserts that
-    ``_prepare_transformation`` can still retrieve the correct transformation.
+    re-assign to sdata_filt — then asserts that ``_prepare_transformation``
+    can still retrieve the correct transformation.
     """
     from spatialdata_plot.pl._datashader import _prepare_transformation
 
@@ -1354,20 +1354,13 @@ def test_groups_filtering_preserves_transformation(sdata_blobs: SpatialData):
     keep = shapes["cluster"] == "c1"
     shapes = shapes[keep].reset_index(drop=True)
     sdata_filt["blobs_polygons"] = shapes
-    # GeoDataFrame re-wrap strips .attrs (this is what _render_shapes does next)
-    shapes = gpd.GeoDataFrame(shapes, geometry="geometry")
 
     # sdata_filt's element must still carry the correct transformation
     trans, _ = _prepare_transformation(sdata_filt.shapes["blobs_polygons"], cs)
     matrix = trans.get_matrix()
     np.testing.assert_allclose(matrix[0, 0], scale_factor, err_msg="x-scale lost after groups filtering")
     np.testing.assert_allclose(matrix[1, 1], scale_factor, err_msg="y-scale lost after groups filtering")
 
-    # The GeoDataFrame re-wrap strips attrs — reading the transform from
-    # the re-wrapped object must fail, proving why early capture matters.
-    with pytest.raises(AssertionError):
-        _prepare_transformation(shapes, cs)
-
 
 def test_plot_can_handle_mixed_numeric_and_color_data(sdata_blobs: SpatialData):
     """Test that mixed numeric and color-like data raises a clear error."""