Fixed categorical coloring of Contours in matplotlib (#2259)

holoviews/core/data/dictionary.py

@@ -210,7 +210,7 @@ def values(cls, dataset, dim, expanded=True, flat=True):
         if np.isscalar(values):
             if not expanded:
                 return np.array([values])
-            values = np.full(len(dataset), values)
+            values = np.full(len(dataset), values, dtype=np.array(values).dtype)
         else:
             if not expanded:
                 return util.unique_array(values)

holoviews/plotting/mpl/path.py

@@ -85,9 +85,12 @@ def get_data(self, element, ranges, style):
             return (paths,), style, {}
 
         if element.level is not None:
-            style['array'] = np.full(len(paths), element.level)
+            array = np.full(len(paths), element.level)
         else:
-            style['array'] = element.dimension_values(cdim, expanded=False)
+            array = element.dimension_values(cdim, expanded=False)
+        if array.dtype.kind not in 'if':
+            array = np.searchsorted(np.unique(array), array)
+        style['array']= array
         self._norm_kwargs(element, ranges, style, cdim)
         style['clim'] = style.pop('vmin'), style.pop('vmax')
         return (paths,), style, {}

tests/testplotinstantiation.py

@@ -341,6 +341,15 @@ def test_image_listed_cmap(self):
         self.assertIsInstance(cmap, ListedColormap)
         self.assertEqual(cmap.colors, colors)
 
+    def test_contours_categorical_color(self):
+        path = Contours([{('x', 'y'): np.random.rand(10, 2), 'z': cat}
+                     for cat in ('B', 'A', 'B')],
+                    vdims='z').opts(plot=dict(color_index='z'))
+        plot = mpl_renderer.get_plot(path)
+        artist = plot.handles['artist']
+        self.assertEqual(artist.get_array(), np.array([1, 0, 1]))
+
+
 
 class TestBokehPlotInstantiation(ComparisonTestCase):