Allow FacetGrid legend to work with nested labels (#1909)

* Add test to reproduce GH#1560

* Allow FacetGrid legend to work with nested labels

Fixes #1560

* Update comments and release notes

doc/releases/v0.9.1.txt

@@ -17,6 +17,8 @@ New features
 
 - It is now possible to force a categorical interpretation of the ``hue`` varaible in a relational plot by passing the name of a categorical palette (e.g. ``"deep"``, or ``"Set2"``). This complements the (previously supported) option of passig a list/dict of colors.
 
+- Added the ability to pass hierarchical label names to the :class:`FacetGrid` legend, which also fixes a bug in :func:`relplot` when the same label appeared in diffent semantics.
+
 Bug fixes and adaptations
 ~~~~~~~~~~~~~~~~~~~~~~~~~
 

seaborn/axisgrid.py

@@ -43,7 +43,8 @@ def add_legend(self, legend_data=None, title=None, label_order=None,
         Parameters
         ----------
         legend_data : dict, optional
-            Dictionary mapping label names to matplotlib artist handles. The
+            Dictionary mapping label names (or two-element tuples where the
+            second element is a label name) to matplotlib artist handles. The
             default reads from ``self._legend_data``.
         title : string, optional
             Title for the legend. The default reads from ``self._hue_var``.
@@ -61,7 +62,8 @@ def add_legend(self, legend_data=None, title=None, label_order=None,
 
         """
         # Find the data for the legend
-        legend_data = self._legend_data if legend_data is None else legend_data
+        if legend_data is None:
+            legend_data = self._legend_data
         if label_order is None:
             if self.hue_names is None:
                 label_order = list(legend_data.keys())
@@ -76,6 +78,15 @@ def add_legend(self, legend_data=None, title=None, label_order=None,
         except TypeError:  # labelsize is something like "large"
             title_size = mpl.rcParams["axes.labelsize"]
 
+        # Unpack nested labels from a hierarchical legend
+        labels = []
+        for entry in label_order:
+            if isinstance(entry, tuple):
+                _, label = entry
+            else:
+                label = entry
+            labels.append(label)
+
         # Set default legend kwargs
         kwargs.setdefault("scatterpoints", 1)
 
@@ -84,7 +95,7 @@ def add_legend(self, legend_data=None, title=None, label_order=None,
             kwargs.setdefault("frameon", False)
 
             # Draw a full-figure legend outside the grid
-            figlegend = self.fig.legend(handles, label_order, "center right",
+            figlegend = self.fig.legend(handles, labels, "center right",
                                         **kwargs)
             self._legend = figlegend
             figlegend.set_title(title, prop={"size": title_size})
@@ -115,7 +126,7 @@ def add_legend(self, legend_data=None, title=None, label_order=None,
         else:
             # Draw a legend in the first axis
             ax = self.axes.flat[0]
-            leg = ax.legend(handles, label_order, loc="best", **kwargs)
+            leg = ax.legend(handles, labels, loc="best", **kwargs)
             leg.set_title(title, prop={"size": title_size})
 
         return self

seaborn/relational.py

@@ -664,8 +664,8 @@ def update(var_name, val_name, **kws):
             artist = func([], [], label=label, **use_kws)
             if self._legend_func == "plot":
                 artist = artist[0]
-            legend_data[label] = artist
-            legend_order.append(label)
+            legend_data[key] = artist
+            legend_order.append(key)
 
         self.legend_data = legend_data
         self.legend_order = legend_order

seaborn/tests/test_axisgrid.py

@@ -256,6 +256,18 @@ def test_get_boolean_legend_data(self):
         for label, level in zip(labels, b_levels):
             nt.assert_equal(label.get_text(), level)
 
+    def test_legend_tuples(self):
+
+        g = ag.FacetGrid(self.df, hue="a")
+        g.map(plt.plot, "x", "y")
+
+        handles, labels = g.ax.get_legend_handles_labels()
+        label_tuples = [("", l) for l in labels]
+        legend_data = dict(zip(label_tuples, handles))
+        g.add_legend(legend_data, label_tuples)
+        for entry, label in zip(g._legend.get_texts(), labels):
+            assert entry.get_text() == label
+
     def test_legend_options(self):
 
         g1 = ag.FacetGrid(self.df, hue="b")

seaborn/tests/test_relational.py

@@ -1706,6 +1706,12 @@ def test_relplot_styles(self, long_df):
             expected_paths = [paths[val] for val in grp_df["a"]]
             assert self.paths_equal(points.get_paths(), expected_paths)
 
+    def test_relplot_stringy_numerics(self, long_df):
+
+        long_df["x_str"] = long_df["x"].astype(str)
+        g = rel.relplot(x="x", y="y", hue="x_str", data=long_df)
+        assert g._legend.texts[0].get_text() == "x_str"
+
     def test_relplot_legend(self, long_df):
 
         g = rel.relplot(x="x", y="y", data=long_df)
@@ -1723,6 +1729,12 @@ def test_relplot_legend(self, long_df):
         g = rel.relplot(x="x", y="y", hue="a", legend=False, data=long_df)
         assert g._legend is None
 
-        long_df["x_str"] = long_df["x"].astype(str)
-        g = rel.relplot(x="x", y="y", hue="x_str", data=long_df)
-        assert g._legend.texts[0].get_text() == "x_str"
+        palette = color_palette("deep", len(long_df["b"].unique()))
+        a_like_b = dict(zip(long_df["a"].unique(), long_df["b"].unique()))
+        long_df["a_like_b"] = long_df["a"].map(a_like_b)
+        g = rel.relplot(x="x", y="y", hue="b", style="a_like_b",
+                        palette=palette, kind="line", estimator=None,
+                        data=long_df)
+        lines = g._legend.get_lines()[1:]  # Chop off title dummy
+        for line, color in zip(lines, palette):
+            assert line.get_color() == color