Implement Widget.from_values (#7033)

doc/how_to/components/index.md

@@ -26,6 +26,13 @@ How to access Pane Type.
 How to access and set widget values.
 :::
 
+:::{grid-item-card} {octicon}`telescope;2.5em;sd-mr-1 sd-animate-grow50` Construct Widget from Data
+:link: widget_from_values
+:link-type: doc
+
+How to automatically infer widget parameters from data values.
+:::
+
 :::{grid-item-card} {octicon}`file-diff;2.5em;sd-mr-1 sd-animate-grow50` Add or Remove Components
 :link: add_remove
 :link-type: doc
@@ -42,5 +49,6 @@ How to add or remove components from a panel.
 construct_panes
 pane_type
 widget_values
+widget_from_values
 add_remove
 ```

doc/how_to/components/widget_from_values.md

@@ -0,0 +1,35 @@
+# Construct Widgets from Data
+
+This guide discusses how to automatically generate widget from data.
+
+---
+
+When working with data, be it in the form of lists, arrays or DataFrames it is common to want to filter that data. Manually computing the `start` and `end` values of a slider or the unique values of a dropdown can be an annoyance so widgets have a `classmethod` called `from_values` to help with this.
+
+```{pyodide}
+import pandas as pd
+import panel as pn
+pn.extension() # for notebook
+
+df = pd.read_csv("https://datasets.holoviz.org/penguins/v1/penguins.csv")
+
+species = pn.widgets.MultiSelect.from_values(df.species)
+
+species
+```
+
+As we can see the special constructor automatically inferred both the `option` and the `name` for the widget.
+
+Similarly we can also use this to infer the values of a numeric column:
+
+```{pyodide}
+body_mass = pn.widgets.RangeSlider.from_values(df.body_mass_g)
+
+body_mass
+```
+
+---
+
+## Related Resources
+
+- Learn about building interactive data pipelines [How-To > Interactivity -> ](../how_to/interactivity/hvplot_interactive.md).

doc/how_to/interactivity/index.md

@@ -32,7 +32,7 @@ Discover how to bind parameters, widgets, and bound functions to components.
 :link: hvplot_interactive
 :link-type: doc
 
-How to use `hvplot.interactive` with widgets to make your data workflows interactive
+How to use `param.rx` with widgets to make your data workflows interactive
 :::
 
 ::::

panel/tests/conftest.py

@@ -277,6 +277,31 @@ def dataframe():
     }, index=[1, 2, 3], columns=['int', 'float', 'str'])
 
 
+@pytest.fixture
+def df_mixed():
+    df = pd.DataFrame({
+        'int': [1, 2, 3, 4],
+        'float': [3.14, 6.28, 9.42, -2.45],
+        'str': ['A', 'B', 'C', 'D'],
+        'bool': [True, True, True, False],
+        'date': [dt.date(2019, 1, 1), dt.date(2020, 1, 1), dt.date(2020, 1, 10), dt.date(2019, 1, 10)],
+        'datetime': [dt.datetime(2019, 1, 1, 10), dt.datetime(2020, 1, 1, 12), dt.datetime(2020, 1, 10, 13), dt.datetime(2020, 1, 15, 13)]
+    }, index=['idx0', 'idx1', 'idx2', 'idx3'])
+    return df
+
+
+@pytest.fixture
+def df_multiindex(df_mixed):
+    df_mi = df_mixed.copy()
+    df_mi.index = pd.MultiIndex.from_tuples([
+        ('group0', 'subgroup0'),
+        ('group0', 'subgroup1'),
+        ('group1', 'subgroup0'),
+        ('group1', 'subgroup1'),
+    ], names=['groups', 'subgroups'])
+    return df_mi
+
+
 @pytest.fixture
 def hv_bokeh():
     import holoviews as hv
@@ -544,18 +569,6 @@ def eh(exception):
         config.exception_handler = old_eh
 
 
-@pytest.fixture
-def df_mixed():
-    df = pd.DataFrame({
-        'int': [1, 2, 3, 4],
-        'float': [3.14, 6.28, 9.42, -2.45],
-        'str': ['A', 'B', 'C', 'D'],
-        'bool': [True, True, True, False],
-        'date': [dt.date(2019, 1, 1), dt.date(2020, 1, 1), dt.date(2020, 1, 10), dt.date(2019, 1, 10)],
-        'datetime': [dt.datetime(2019, 1, 1, 10), dt.datetime(2020, 1, 1, 12), dt.datetime(2020, 1, 10, 13), dt.datetime(2020, 1, 15, 13)]
-    }, index=['idx0', 'idx1', 'idx2', 'idx3'])
-    return df
-
 @pytest.fixture
 def df_strings():
     descr = [

panel/tests/ui/widgets/test_tabulator.py

@@ -71,18 +71,6 @@ def df_mixed_as_string():
     """
 
 
-@pytest.fixture
-def df_multiindex(df_mixed):
-    df_mi = df_mixed.copy()
-    df_mi.index = pd.MultiIndex.from_tuples([
-        ('group0', 'subgroup0'),
-        ('group0', 'subgroup1'),
-        ('group1', 'subgroup0'),
-        ('group1', 'subgroup1'),
-    ], names=['groups', 'subgroups'])
-    return df_mi
-
-
 def count_per_page(count: int, page_size: int):
     """
     >>> count_per_page(12, 7)

panel/tests/widgets/test_select.py

@@ -1,4 +1,5 @@
 import numpy as np
+import pandas as pd
 import pytest
 
 from panel.layout import GridBox, Row
@@ -39,6 +40,32 @@ def test_select_text_option_with_equality(widget):
     select.value = 'DEF'
     assert select.value == 'DEF'
 
+def test_select_from_list(document, comm):
+    select = Select.from_values(['A', 'B', 'A', 'B', 'C'])
+
+    assert select.options == ['A', 'B', 'C']
+    assert select.value == 'A'
+
+def test_select_from_array(document, comm):
+    select = Select.from_values(np.array(['A', 'B', 'A', 'B', 'C']))
+
+    assert select.options == ['A', 'B', 'C']
+    assert select.value == 'A'
+
+def test_select_from_index(document, comm):
+    select = Select.from_values(pd.Index(['A', 'B', 'A', 'B', 'C'], name='index'))
+
+    assert select.options == ['A', 'B', 'C']
+    assert select.value == 'A'
+    assert select.name == 'index'
+
+def test_select_from_series(document, comm):
+    select = Select.from_values(pd.Series(['A', 'B', 'A', 'B', 'C'], name='Series'))
+
+    assert select.options == ['A', 'B', 'C']
+    assert select.value == 'A'
+    assert select.name == 'Series'
+
 def test_select(document, comm):
     opts = {'A': 'a', '1': 1}
     select = Select(options=opts, value=opts['1'], name='Select')
@@ -275,6 +302,31 @@ def test_nested_select_defaults(document, comm):
     assert select._max_depth == 3
 
 
+def test_nested_select_from_multi_index(df_multiindex):
+    select = NestedSelect.from_values(df_multiindex.index)
+
+    assert select.options == {
+        'group0': ['subgroup0', 'subgroup1'],
+        'group1': ['subgroup0', 'subgroup1'],
+    }
+    assert select.value == {'groups': 'group0', 'subgroups': 'subgroup0'}
+    assert select._max_depth == 2
+    assert select.levels == ['groups', 'subgroups']
+
+def test_nested_select_from_index():
+    select = NestedSelect.from_values(pd.Index(['A', 'B', 'A', 'B', 'C'], name='index'))
+
+    assert select.options == ['A', 'B', 'C']
+    assert select.value == {'index': 'A'}
+    assert select._max_depth == 1
+
+def test_nested_select_from_series():
+    select = NestedSelect.from_values(pd.Series(['A', 'B', 'A', 'B', 'C'], name='Series'))
+
+    assert select.options == ['A', 'B', 'C']
+    assert select.value == {'Series': 'A'}
+    assert select._max_depth == 1
+
 def test_nested_select_init_value(document, comm):
     options = {
         "Andrew": {
@@ -856,6 +908,32 @@ def test_select_disabled_options_set_value_and_disabled_options(options, size, d
     assert widget.disabled_options == [10]
 
 
+def test_multi_select_from_list():
+    select = MultiSelect.from_values(['A', 'B', 'A', 'B', 'C'])
+
+    assert select.options == ['A', 'B', 'C']
+    assert select.value == []
+
+def test_multi_select_from_array():
+    select = MultiSelect.from_values(np.array(['A', 'B', 'A', 'B', 'C']))
+
+    assert select.options == ['A', 'B', 'C']
+    assert select.value == []
+
+def test_multi_select_from_index():
+    select = MultiSelect.from_values(pd.Index(['A', 'B', 'A', 'B', 'C'], name='index'))
+
+    assert select.options == ['A', 'B', 'C']
+    assert select.value == []
+    assert select.name == 'index'
+
+def test_multi_select_from_series(document, comm):
+    select = MultiSelect.from_values(pd.Series(['A', 'B', 'A', 'B', 'C'], name='Series'))
+
+    assert select.options == ['A', 'B', 'C']
+    assert select.value == []
+    assert select.name == 'Series'
+
 def test_multi_select(document, comm):
     select = MultiSelect(options={'A': 'A', '1': 1, 'C': object},
                          value=[object, 1], name='Select')

panel/tests/widgets/test_slider.py

@@ -1,6 +1,7 @@
 from datetime import date, datetime
 
 import numpy as np
+import pandas as pd
 import pytest
 
 from bokeh.models import (
@@ -15,6 +16,28 @@
 )
 
 
+def test_float_slider_from_list():
+    slider = FloatSlider.from_values([1.1, 2.2])
+
+    assert slider.start == 1.1
+    assert slider.end == 2.2
+    assert slider.value == 1.1
+
+def test_float_slider_from_array():
+    slider = FloatSlider.from_values(np.array([1.1, 2.2]))
+
+    assert slider.start == 1.1
+    assert slider.end == 2.2
+    assert slider.value == 1.1
+
+def test_float_slider_from_series():
+    slider = FloatSlider.from_values(pd.Series([1.1, 2.2], name='Series'))
+
+    assert slider.start == 1.1
+    assert slider.end == 2.2
+    assert slider.value == 1.1
+    assert slider.name == 'Series'
+
 def test_float_slider(document, comm):
 
     slider = FloatSlider(start=0.1, end=0.5, value=0.4, name='Slider')
@@ -84,6 +107,29 @@ def test_int_slider(document, comm):
         assert widget.value == 2
 
 
+
+def test_range_slider_from_list():
+    slider = RangeSlider.from_values([1.1, 2.2])
+
+    assert slider.start == 1.1
+    assert slider.end == 2.2
+    assert slider.value == (1.1, 2.2)
+
+def test_range_slider_from_array():
+    slider = RangeSlider.from_values(np.array([1.1, 2.2]))
+
+    assert slider.start == 1.1
+    assert slider.end == 2.2
+    assert slider.value == (1.1, 2.2)
+
+def test_range_slider_from_series():
+    slider = RangeSlider.from_values(pd.Series([1.1, 2.2], name='Series'))
+
+    assert slider.start == 1.1
+    assert slider.end == 2.2
+    assert slider.value == (1.1, 2.2)
+    assert slider.name == 'Series'
+
 def test_range_slider(document, comm):
 
     slider = RangeSlider(start=0., end=3, value=(0, 3), name='Slider')

panel/util/__init__.py

@@ -508,6 +508,16 @@ def safe_next():
             break
         yield value
 
+def unique_iterator(seq):
+    """
+    Returns an iterator containing all non-duplicate elements
+    in the input sequence.
+    """
+    seen = set()
+    for item in seq:
+        if item not in seen:
+            seen.add(item)
+            yield item
 
 def prefix_length(a: str, b: str) -> int:
     """

panel/widgets/base.py

@@ -12,6 +12,7 @@
     TYPE_CHECKING, Any, ClassVar, TypeVar,
 )
 
+import numpy as np
 import param  # type: ignore
 
 from bokeh.models import ImportedStyleSheet, Tooltip
@@ -21,6 +22,7 @@
 from .._param import Margin
 from ..layout.base import Row
 from ..reactive import Reactive
+from ..util import unique_iterator
 from ..viewable import Layoutable, Viewable
 
 if TYPE_CHECKING:
@@ -69,6 +71,45 @@ def from_param(cls: type[T], parameter: param.Parameter, **params) -> T:
         )
         return layout[0]
 
+    @classmethod
+    def _infer_params(cls, values, **params):
+        if 'name' not in params and getattr(values, 'name', None):
+            params['name'] = values.name
+        if 'start' in cls.param and 'start' not in params:
+            params['start'] = np.nanmin(values)
+        if 'end' in cls.param and 'end' not in params:
+            params['end'] = np.nanmax(values)
+        if 'options' in cls.param and 'options' not in params:
+            if isinstance(values, dict):
+                params['options'] = values
+            else:
+                params['options'] = list(unique_iterator(values))
+        if 'value' not in params:
+            p = cls.param['value']
+            if isinstance(p, param.Tuple):
+                params['value'] = (params['start'], params['end'])
+            elif 'start' in params:
+                params['value'] = params['start']
+            elif ('options' in params and not isinstance(p, (param.List, param.ListSelector))
+                  and not getattr(cls, '_allows_none', False)):
+                params['value'] = params['options'][0]
+        return params
+
+    @classmethod
+    def from_values(cls, values, **params):
+        """
+        Creates an instance of this Widget where the parameters are
+        inferred from the data.
+
+        Arguments
+        ---------
+        values: Iterable
+            The values to infer the parameters from.
+        params: dict
+            Additional parameters to pass to the widget.
+        """
+        return cls(**cls._infer_params(values, **params))
+
     @property
     def rx(self):
         return self.param.value.rx