Add str support to dropdown (#529)

This small PR adds str column support to `drop_down` and `multi_select` widgets, along with `charts.bar`, which already supports str columns directly since 23.08

```python
import cudf
from cuxfilter import charts, DataFrame

cux_df = DataFrame.from_dataframe(cudf.DataFrame({'val': ['A', 'B', 'C', 'D', 'E']}))
multi_select = charts.multi_select('val')

d = cux_df.dashboard([multi_select])
# View the individual multi_select chart part of the dashboard d
multi_select.view(height=200)
```
![image](https://github.com/rapidsai/cuxfilter/assets/20476096/5768f5ca-8915-410f-9d10-065b89cc94df)

cc @exactlyallan

Authors:
  - Ajay Thorve (https://github.com/AjayThorve)

Approvers:
  - Allan (https://github.com/exactlyallan)

URL: #529

docs/source/user_guide/charts/widgets.rst

@@ -99,12 +99,12 @@ Example
     import cudf
     from cuxfilter import charts, DataFrame
 
-    cux_df = DataFrame.from_dataframe(cudf.DataFrame({'key': [0, 1, 2, 3, 4], 'val':[float(i + 10) for i in range(5)]}))
-    drop_down = charts.drop_down('val')
+    cux_df = DataFrame.from_dataframe(cudf.DataFrame({'val': ['A', 'B', 'C', 'D', 'E']}))
+    multi_select = charts.drop_down('val')
 
-    d = cux_df.dashboard([drop_down])
-    #view the individual drop_down chart part of the dashboard d
-    drop_down.view()
+    d = cux_df.dashboard([multi_select])
+    # View the individual multi_select chart part of the dashboard d
+    multi_select.view(height=200)
 
 
 Multiselect
@@ -118,12 +118,12 @@ Example
     import cudf
     from cuxfilter import charts, DataFrame
 
-    cux_df = DataFrame.from_dataframe(cudf.DataFrame({'key': [0, 1, 2, 3, 4], 'val':[float(i + 10) for i in range(5)]}))
+    cux_df = DataFrame.from_dataframe(cudf.DataFrame({'val': ['A', 'B', 'C', 'D', 'E']}))
     multi_select = charts.multi_select('val')
 
     d = cux_df.dashboard([multi_select])
-    #view the individual multi_select chart part of the dashboard d
-    multi_select.view()
+    # View the individual multi_select chart part of the dashboard d
+    multi_select.view(height=200)
 
 
 Number Chart

python/cuxfilter/charts/core/core_widget.py

@@ -110,7 +110,7 @@ def _repr_mimebundle_(self, include=None, exclude=None):
         return None
 
     def view(self, width=400, height=10):
-        return pn.panel(self.chart, width=width, height=10)
+        return pn.Column(self.chart, width=width, height=height)
 
     def get_dashboard_view(self):
         return pn.panel(self.chart, sizing_mode="stretch_width")

python/cuxfilter/charts/core/non_aggregate/core_non_aggregate.py

@@ -106,7 +106,6 @@ def cb(bounds, x_selection, y_selection):
                 dashboard_cls._query_str_dict,
                 dashboard_cls._query_local_variables_dict,
             )
-            print("hello")
             # reload all charts with new queried data (cudf.DataFrame only)
             dashboard_cls._reload_charts()
 

python/cuxfilter/charts/panel_widgets/panel_widgets.py

@@ -28,7 +28,7 @@ def range_slider(
     ----------
 
     x: str
-        column name from gpu dataframe
+        column name from gpu dataframe, dtype should be int or float
 
     data_points: int,  default None
         when None, it means no custom number of bins are provided and
@@ -64,7 +64,7 @@ def date_range_slider(
     ----------
 
     x: str
-        column name from gpu dataframe
+        column name from gpu dataframe, dtype should be datetime
 
     data_points: int,  default None
         when None, it means no custom number of bins are provided and
@@ -99,7 +99,7 @@ def int_slider(x, data_points=None, step_size=1, **params):
     ----------
 
     x: str
-        column name from gpu dataframe
+        column name from gpu dataframe, dtype should be int
 
     data_points: int,  default None
         when None, it means no custom number of bins are provided and
@@ -129,7 +129,7 @@ def float_slider(x, data_points=None, step_size=None, **params):
     ----------
 
     x: str
-        column name from gpu dataframe
+        column name from gpu dataframe, dtype should be float
 
     data_points: int,  default None
         when None, it means no custom number of bins are provided and
@@ -165,7 +165,7 @@ def drop_down(x, **params):
     ----------
 
     x: str
-        column name from gpu dataframe
+        column name from gpu dataframe, dtype [str, int, float]
 
     data_points: int,  default number of unique values
 
@@ -190,7 +190,7 @@ def multi_select(x, **params):
     ----------
 
     x: str
-        column name from gpu dataframe
+        column name from gpu dataframe, dtype [str, int, float]
 
     data_points: int,  default number of unique values
 

python/cuxfilter/charts/panel_widgets/plots.py

@@ -319,7 +319,7 @@ def compute_query_dict(self, query_str_dict, query_local_variables_dict):
 
 
 class MultiChoice(BaseWidget):
-    value = None
+    source = None
 
     def initiate_chart(self, dashboard_cls):
         """
@@ -328,16 +328,11 @@ def initiate_chart(self, dashboard_cls):
         self.min_value, self.max_value = get_min_max(
             dashboard_cls._cuxfilter_df.data, self.x
         )
-
-        if self.stride is None:
-            if self.max_value < 1 and self.stride_type == int:
-                self.stride_type = float
-            self.stride = self.stride_type(1)
-
+        self.source = dashboard_cls._cuxfilter_df.data[self.x].reset_index(
+            drop=True
+        )
         self.calc_list_of_values(dashboard_cls._cuxfilter_df.data)
-
         self.generate_widget()
-
         self.add_events(dashboard_cls)
 
     def calc_list_of_values(self, data):
@@ -398,13 +393,21 @@ def compute_query_dict(self, query_str_dict, query_local_variables_dict):
         query_dict:
             reference to dashboard.__cls__.query_dict
         """
-        if len(self.chart.value) == 0 or self.chart.value == [""]:
+        if len(self.chart.value) == 0:
             query_str_dict.pop(self.name, None)
-        elif len(self.chart.value) == 1:
-            query_str_dict[self.name] = f"{self.x}=={self.chart.value[0]}"
         else:
-            indices_string = ",".join(map(str, self.chart.value))
-            query_str_dict[self.name] = f"{self.x} in ({indices_string})"
+            df_module = (
+                cudf if isinstance(self.source, cudf.Series) else dask_cudf
+            )
+
+            if self.source.dtype == "object":
+                query_str_dict[self.name] = df_module.DataFrame(
+                    self.source.str.contains("|".join(self.chart.value))
+                )
+            else:
+                query_str_dict[self.name] = df_module.DataFrame(
+                    self.source.isin(self.chart.value)
+                )
 
     def apply_theme(self, theme):
         """

python/cuxfilter/tests/charts/core/test_core_widget.py

@@ -31,7 +31,7 @@ def test_view(self, chart, _chart):
         bw = BaseWidget("test_x")
         bw.chart = chart
 
-        assert str(bw.view()) == str(pn.panel(_chart, width=400, height=10))
+        assert str(bw.view()) == str(pn.Column(_chart, width=400, height=10))
 
     def test_add_event(self):
         bw = BaseWidget("test_x")