Using Tabulator¶

Tabulator is the preferred widget for displaying DataFrames in Panel apps — sortable, filterable, paginated. Requires pn.extension("tabulator").

Basic Setup¶

self._table = pn.widgets.Tabulator(
    value=df,
    sizing_mode="stretch_width",
    layout="fit_columns",
    disabled=True,            # prevent cell editing
    show_index=False,
    theme="materialize",
    pagination="remote",
    page_size=15,
    formatters={
        "Revenue": {"type": "money", "symbol": "$", "precision": 0},
        "Cost": {"type": "money", "symbol": "$", "precision": 0},
    },
)

layout="fit_columns" fills available width. Best option for responsive layouts.
disabled=True prevents editing — set False for editable tables.
Prefer Tabulator formatters/editors over Bokeh types.

Widget-Based Filtering with `add_filter`¶

Wire sidebar widgets directly as table filters — Tabulator watches widget values and shows/hides rows automatically. No manual filtering needed.

region_filter = pmui.CheckBoxGroup(value=list(REGIONS), options=REGIONS)
product_filter = pmui.CheckBoxGroup(value=list(PRODUCTS), options=PRODUCTS)

# Must be called AFTER table has data (value is set),
# otherwise raises AttributeError: 'NoneType' has no attribute 'columns'
table.add_filter(region_filter, "Region")
table.add_filter(product_filter, "Product")

Filter behavior depends on the widget value type: a list matches any item in the list, a tuple is interpreted as a range (inclusive), a scalar checks for equality.

Use table.current_view to inspect the currently visible (filtered + sorted) DataFrame.

Function-Based Filtering¶

For complex filtering, pass a bound function:

search = pn.widgets.TextInput(name="Search", value="")

def contains_filter(df, pattern, column):
    if not pattern:
        return df
    return df[df[column].str.contains(pattern, case=False)]

table.add_filter(pn.bind(contains_filter, pattern=search, column="Name"))

Checkbox Selection¶

Use selectable="checkbox" for row selection with checkboxes. Watch the selection parameter for changes and read selected_dataframe for the selected rows.

self._table = pn.widgets.Tabulator(
    ...,
    selectable="checkbox",
)
self._table.param.watch(self._on_selection, "selection")

def _on_selection(self, event):
    selected = self._table.selected_dataframe
    if selected is not None and not selected.empty:
        # Use selected rows to filter charts, update KPIs, etc.
        ...

Row Content (Expandable Details)¶

Use row_content to render an expandable detail region below each row. Pass a function that receives a pandas.Series (the row) and returns a Panel object.

def row_detail(row):
    return pn.Column(
        f"**{row['Name']}** — {row['Region']}",
        pn.pane.HoloViews(
            df[df["Name"] == row["Name"]].hvplot.line(x="date", y="revenue"),
            height=200,
        ),
    )

self._table = pn.widgets.Tabulator(
    value=df,
    row_content=row_detail,
    sizing_mode="stretch_width",
)

The function is called lazily when the user expands a row (clicks the + icon).
Set embed_content=True to pre-render all row content on load instead of lazily — useful when there are few rows and you want instant expansion, but expensive with many rows.