Support interchange protocol

ci/deps_pinned.txt

@@ -1,5 +1,7 @@
 numpy~=1.20.0
 pandas~=1.2.0
+polars~=0.17.0
+pyarrow~=12.0.0
 matplotlib~=3.3.0
 scipy~=1.7.0
 statsmodels~=0.12.0

seaborn/_core/data.py

@@ -10,6 +10,7 @@
 from pandas import DataFrame
 
 from seaborn._core.typing import DataSource, VariableSpec, ColumnName
+from seaborn.utils import try_convert_to_pandas
 
 
 class PlotData:
@@ -51,7 +52,7 @@ def __init__(
         data: DataSource,
         variables: dict[str, VariableSpec],
     ):
-
+        data = try_convert_to_pandas(data)
         frame, names, ids = self._assign_variables(data, variables)
 
         self.frame = frame

seaborn/_core/plot.py

@@ -42,7 +42,7 @@
 from seaborn._compat import set_scale_obj, set_layout_engine
 from seaborn.rcmod import axes_style, plotting_context
 from seaborn.palettes import color_palette
-from seaborn.utils import _version_predates
+from seaborn.utils import _version_predates, try_convert_to_pandas
 
 from typing import TYPE_CHECKING, TypedDict
 if TYPE_CHECKING:
@@ -309,6 +309,7 @@ def __init__(
 
         if args:
             data, variables = self._resolve_positionals(args, data, variables)
+        data = try_convert_to_pandas(data)
 
         unknown = [x for x in variables if x not in PROPERTIES]
         if unknown:
@@ -347,7 +348,10 @@ def _resolve_positionals(
 
         # TODO need some clearer way to differentiate data / vector here
         # (There might be an abstract DataFrame class to use here?)
-        if isinstance(args[0], (abc.Mapping, pd.DataFrame)):
+        if (
+            isinstance(args[0], (abc.Mapping, pd.DataFrame))
+            or hasattr(args[0], '__dataframe__')
+        ):
             if data is not None:
                 raise TypeError("`data` given by both name and position.")
             data, args = args[0], args[1:]

seaborn/_oldcore.py

@@ -775,7 +775,7 @@ def _assign_variables_wideform(self, data=None, **kwargs):
             # (Could be accomplished with a more general to_series() interface)
             flat_data = pd.Series(data).copy()
             names = {
-                "@values": flat_data.name,
+                "@values": getattr(data, 'name', None),
                 "@index": flat_data.index.name
             }
 
@@ -922,7 +922,7 @@ def _assign_variables_longform(self, data=None, **kwargs):
                     val in data
                     or (isinstance(val, (str, bytes)) and val in index)
                 )
-            except (KeyError, TypeError):
+            except (KeyError, TypeError, ValueError):
                 val_as_data_key = False
 
             if val_as_data_key:

seaborn/axisgrid.py

@@ -372,7 +372,7 @@ def __init__(
         margin_titles=False, xlim=None, ylim=None, subplot_kws=None,
         gridspec_kws=None,
     ):
-
+        data = utils.try_convert_to_pandas(data)
         super().__init__()
 
         # Determine the hue facet layer information
@@ -1238,7 +1238,7 @@ def __init__(
         .. include:: ../docstrings/PairGrid.rst
 
         """
-
+        data = utils.try_convert_to_pandas(data)
         super().__init__()
 
         # Sort out the variables that define the grid
@@ -2087,6 +2087,8 @@ def pairplot(
     # Avoid circular import
     from .distributions import histplot, kdeplot
 
+    data = utils.try_convert_to_pandas(data)
+
     # Handle deprecations
     if size is not None:
         height = size

seaborn/categorical.py

@@ -2490,7 +2490,6 @@ def stripplot(
     hue_norm=None, native_scale=False, formatter=None, legend="auto",
     ax=None, **kwargs
 ):
-
     p = _CategoricalPlotterNew(
         data=data,
         variables=_CategoricalPlotterNew.get_semantics(locals()),
@@ -2618,6 +2617,7 @@ def swarmplot(
     ax=None, **kwargs
 ):
 
+    data = utils.try_convert_to_pandas(data)
     p = _CategoricalPlotterNew(
         data=data,
         variables=_CategoricalPlotterNew.get_semantics(locals()),
@@ -3162,6 +3162,7 @@ def catplot(
     margin_titles=False, facet_kws=None, ci="deprecated",
     **kwargs
 ):
+    data = utils.try_convert_to_pandas(data)
 
     # Determine the plotting function
     try:

seaborn/distributions.py

@@ -33,6 +33,7 @@
     _check_argument,
     _assign_default_kwargs,
     _default_color,
+    try_convert_to_pandas,
 )
 from .palettes import color_palette
 from .external import husl
@@ -1392,6 +1393,7 @@ def histplot(
     # Other appearance keywords
     **kwargs,
 ):
+    data = try_convert_to_pandas(data)
 
     p = _DistributionPlotter(
         data=data,
@@ -2123,6 +2125,7 @@ def displot(
     **kwargs,
 ):
 
+    data = try_convert_to_pandas(data)
     p = _DistributionFacetPlotter(
         data=data,
         variables=_DistributionFacetPlotter.get_semantics(locals())

seaborn/regression.py

@@ -575,6 +575,7 @@ def lmplot(
     truncate=True, x_jitter=None, y_jitter=None, scatter_kws=None,
     line_kws=None, facet_kws=None,
 ):
+    data = utils.try_convert_to_pandas(data)
 
     if facet_kws is None:
         facet_kws = {}

seaborn/relational.py

@@ -13,6 +13,7 @@
     adjust_legend_subtitles,
     _default_color,
     _deprecate_ci,
+    try_convert_to_pandas,
 )
 from ._statistics import EstimateAggregator
 from .axisgrid import FacetGrid, _facet_docs
@@ -799,7 +800,7 @@ def relplot(
     legend="auto", kind="scatter", height=5, aspect=1, facet_kws=None,
     **kwargs
 ):
-
+    data = try_convert_to_pandas(data)
     if kind == "scatter":
 
         plotter = _ScatterPlotter

seaborn/utils.py

@@ -1,4 +1,6 @@
 """Utility functions, mostly for internal use."""
+from __future__ import annotations
+
 import os
 import inspect
 import warnings
@@ -894,3 +896,19 @@ def _disable_autolayout():
 def _version_predates(lib: ModuleType, version: str) -> bool:
     """Helper function for checking version compatibility."""
     return Version(lib.__version__) < Version(version)
+
+
+def try_convert_to_pandas(data: object | None) -> pd.DataFrame:
+    if data is None:
+        return None
+    elif isinstance(data, pd.DataFrame):
+        return data
+    elif hasattr(data, "__dataframe__") and _version_predates(pd, "2.0.2"):
+        msg = (
+            "Interchanging to pandas requires at least pandas version '2.0.2'. "
+            "Please upgrade pandas to at least version '2.0.2'."
+        )
+        raise RuntimeError(msg)
+    elif hasattr(data, "__dataframe__"):
+        return pd.api.interchange.from_dataframe(data)
+    return data

tests/test_utils.py

@@ -433,6 +433,18 @@ def test_move_legend_input_checks():
 def check_load_dataset(name):
     ds = load_dataset(name, cache=False)
     assert isinstance(ds, pd.DataFrame)
+    # Check that the example datasets can actually be interchanged.
+    try:
+        import polars as pl
+    except ModuleNotFoundError:
+        pass
+    else:
+        if _version_predates(pd, '2.0.2'):
+            with pytest.raises(RuntimeError, match='Please upgrade pandas'):
+                utils.try_convert_to_pandas(pl.from_pandas(ds))
+        else:
+            ds = utils.try_convert_to_pandas(pl.from_pandas(ds))
+            assert isinstance(ds, pd.DataFrame)
 
 
 def check_load_cached_dataset(name):