Merge pull request #652 from blooop/feature/tabulator_result

Feature/tabulator result

Author: blooop

bencher/example/inputs_0_float/example_3_cat_in_2_out.py

@@ -103,9 +103,15 @@ def example_3_cat_in_2_out(
         - Note that variance in the results simulates real-world measurement fluctuations
         """,
     )
+
+    res = bench.get_result()
+
+    bench.report.append(res.to(bch.BarResult, agg_over_dims=["data_structure", "data_size"]))
+    bench.report.append(res.to(bch.BarResult, agg_over_dims=["data_structure"]))
+    bench.report.append(res.to(bch.BarResult, agg_over_dims=["data_size"]))
     return bench
 
 
 if __name__ == "__main__":
     br = bch.BenchRunner()
-    br.add(example_3_cat_in_2_out).run(repeats=5, show=True)
+    br.add(example_3_cat_in_2_out).run(repeats=1, show=True)

bencher/example/inputs_2_float/example_2_float_1_cat_in_2_out.py

@@ -91,7 +91,7 @@ def example_2_float_1_cat_in_2_out(
     """
     if run_cfg is None:
         run_cfg = bch.BenchRunCfg()
-    run_cfg.repeats = 3  # Fewer repeats for a quicker benchmark
+    run_cfg.repeats = 1  # Fewer repeats for a quicker benchmark
 
     bench = Pattern1CatBenchmark().to_bench(run_cfg, report)
     bench.plot_sweep(
@@ -108,6 +108,20 @@ def example_2_float_1_cat_in_2_out(
     res = bench.get_result()
 
     bench.report.append(res.to(bch.HeatmapResult, agg_over_dims=["pattern_type"]))
+    # bench.report.append(res.to(bch.TabulatorResult, agg_over_dims=["pattern_type", "x_value"]))
+    # bench.report.append(res.to(bch.TableResult, agg_over_dims=["pattern_type", "x_value", "y_value"]))
+
+    bench.report.append(
+        res.to(bch.TabulatorResult, agg_over_dims=["pattern_type", "x_value", "y_value"])
+    )
+    bench.report.append(
+        res.to(
+            bch.TabulatorResult,
+            agg_over_dims=["pattern_type", "x_value"],
+        )
+    )
+    bench.report.append(res.to(bch.TabulatorResult, agg_over_dims=["pattern_type"]))
+    bench.report.append(res.to(bch.TabulatorResult))
 
     return bench
 

bencher/results/holoview_results/bar_result.py

@@ -90,11 +90,25 @@ def to_bar_ds(self, dataset: xr.Dataset, result_var: Parameter = None, **kwargs)
         Returns:
             hvplot.element.Bars: A bar chart visualization of the benchmark data.
         """
-        by = None
-        if self.plt_cnt_cfg.cat_cnt >= 2 and self.plt_cnt_cfg.cat_vars[1].name in dataset.dims:
-            by = self.plt_cnt_cfg.cat_vars[1].name
-
+        # Determine grouping ('by') dynamically based on dims that still exist
         da = dataset[result_var.name]
+
+        # Allow explicit override via kwargs
+        by = kwargs.pop("by", None)
+        if by is None:
+            # Candidate categorical dims from the original config, filtered to those still present
+            cat_dim_names = [cv.name for cv in self.plt_cnt_cfg.cat_vars]
+            dims_present = [d for d in da.dims if d not in ("repeat", "over_time")]
+            # Prefer categorical dims that are not the primary x-axis
+            candidates = [d for d in dims_present if d != da.dims[0] and d in cat_dim_names]
+
+            if len(candidates) == 1:
+                by = candidates[0]
+            elif len(candidates) > 1:
+                # Preserve multi-level grouping when multiple categorical dims remain
+                by = candidates
+            else:
+                by = None
         title = self.title_from_ds(da, result_var, **kwargs)
         time_args = self.time_widget(title)
 

bencher/results/holoview_results/tabulator_result.py

@@ -3,6 +3,11 @@
 
 from bencher.results.holoview_results.holoview_result import HoloviewResult
 
+from param import Parameter
+import hvplot.xarray  # noqa pylint: disable=duplicate-code,unused-import
+import xarray as xr
+import pandas as pd
+
 
 class TabulatorResult(HoloviewResult):
     def to_plot(self, **kwargs) -> pn.widgets.Tabulator:  # pylint:disable=unused-argument
@@ -17,4 +22,54 @@ def to_plot(self, **kwargs) -> pn.widgets.Tabulator:  # pylint:disable=unused-ar
         Returns:
             pn.widgets.Tabulator: An interactive table widget.
         """
-        return pn.widgets.Tabulator(self.to_pandas())
+        return self.to_tabulator(**kwargs)
+
+    def to_tabulator(self, result_var: Parameter = None, **kwargs) -> pn.widgets.Tabulator:
+        """Generates a Tabulator widget from benchmark data.
+
+        This is a convenience method that calls to_tabulator_ds() with the same parameters.
+
+        Args:
+            result_var (Parameter, optional): The result variable to include in the table. If None, uses the default.
+            **kwargs: Additional keyword arguments passed to the Tabulator constructor.
+
+        Returns:
+            pn.widgets.Tabulator: An interactive table widget.
+        """
+        return self.filter(
+            self.to_tabulator_ds,
+            result_var=result_var,
+            **kwargs,
+        )
+
+    def to_tabulator_ds(
+        self, dataset: xr.Dataset, result_var: Parameter, **kwargs
+    ) -> pn.widgets.Tabulator:
+        """Creates a Tabulator widget from the provided dataset.
+
+        Given a filtered dataset, this method generates an interactive table visualization.
+
+        Args:
+            dataset (xr.Dataset): The filtered dataset to visualize.
+            result_var (Parameter): The result variable to include in the table.
+            **kwargs: Additional keyword arguments passed to the Tabulator constructor.
+
+        Returns:
+            pn.widgets.Tabulator: An interactive table widget.
+        """
+
+        # Assume input is an xarray.Dataset. Keep Dataset throughout.
+        ds: xr.Dataset = dataset if isinstance(dataset, xr.Dataset) else xr.Dataset(dataset)
+
+        # Step 1: If a result variable is specified, select it and keep as Dataset
+        if result_var is not None and result_var.name in ds.data_vars:
+            ds = ds[[result_var.name]]
+
+        # Step 2: Build a flat pandas DataFrame
+        if len(ds.dims) == 0:
+            df = pd.DataFrame({name: [da.values.item()] for name, da in ds.data_vars.items()})
+        else:
+            # N-D: to DataFrame and reset the index so coordinates become columns
+            df = ds.to_dataframe().reset_index()
+
+        return pn.widgets.Tabulator(df, **kwargs)