Data Explorer: More gracefully handle summary statistics for all-null…

… arrays for pandas and polars (#4329)

As described in #4307, for pandas we were raising a warning for all null
/ NA arrays when computing summary stats. The polars implementation
turned out to have a different bug, so that is fixed and tested here
also.

### QA Notes

Follow example reported in #4307 for both pandas and polars (substitute
all None values for polars)

extensions/positron-python/python_files/positron/positron_ipykernel/data_explorer.py

@@ -930,42 +930,33 @@ def _pandas_datetimetz_mapper(type_name):
 def _pandas_summarize_number(col: "pd.Series", options: FormatOptions):
     float_format = _get_float_formatter(options)
 
+    min_val = max_val = median_val = mean_val = std_val = None
     if "complex" in str(col.dtype):
-        min_val = max_val = median_val = mean_val = None
         values = col.to_numpy()
         non_null_values = values[~np_.isnan(values)]
         if len(non_null_values) > 0:
             median_val = float_format(np_.median(non_null_values))
             mean_val = float_format(np_.mean(non_null_values))
-        return _box_number_stats(
-            min_val,
-            max_val,
-            mean_val,
-            median_val,
-            None,  # std_val
-        )
     else:
-        return _get_float_stats_generic(col, float_format)
-
+        non_null_values = col[col.notna()].to_numpy()  # type: ignore
 
-def _get_float_stats_generic(col, float_format):
-    # This works for both pandas and polars because their API for
-    # these functions is the same
-    median_val = mean_val = std_val = None
+        if len(non_null_values) > 0:
+            min_val = non_null_values.min()
+            max_val = non_null_values.max()
 
-    min_val = col.min()
-    max_val = col.max()
+            if not _isinf(min_val) and not _isinf(max_val):
+                # These stats are not defined when there is an
+                # inf/-inf in the data
+                mean_val = float_format(non_null_values.mean())
+                median_val = float_format(np_.median(non_null_values))
+                std_val = float_format(non_null_values.std(ddof=1))
 
-    if not _isinf(min_val) and not _isinf(max_val):
-        # These stats are not defined when there is an
-        # inf/-inf in the data
-        mean_val = float_format(col.mean())
-        median_val = float_format(col.median())
-        std_val = float_format(col.std())
+            min_val = float_format(min_val)
+            max_val = float_format(max_val)
 
     return _box_number_stats(
-        float_format(min_val),
-        float_format(max_val),
+        min_val,
+        max_val,
         mean_val,
         median_val,
         std_val,
@@ -1839,7 +1830,31 @@ def _parse_iso8601_like(x, tz=None):
 
 def _polars_summarize_number(col: "pl.Series", options: FormatOptions):
     float_format = _get_float_formatter(options)
-    return _get_float_stats_generic(col, float_format)
+    min_val = max_val = median_val = mean_val = std_val = None
+
+    is_empty = col.null_count() == len(col)
+
+    if not is_empty:
+        min_val = col.min()
+        max_val = col.max()
+
+        if not _isinf(min_val) and not _isinf(max_val):  # type: ignore
+            # These stats are not defined when there is an
+            # inf/-inf in the data
+            mean_val = float_format(col.mean())
+            median_val = float_format(col.median())
+            std_val = float_format(col.std())
+
+        min_val = float_format(min_val)
+        max_val = float_format(max_val)
+
+    return _box_number_stats(
+        min_val,
+        max_val,
+        mean_val,
+        median_val,
+        std_val,
+    )
 
 
 def _polars_summarize_string(col: "pl.Series", _):

extensions/positron-python/python_files/positron/positron_ipykernel/tests/test_data_explorer.py

@@ -2416,6 +2416,7 @@ def test_pandas_profile_summary_stats(dxf: DataExplorerFixture):
             ),  # datetime single tz
             "f7": [1 + 1j, 2 + 2j, 3 + 3j, 4 + 4j, np.nan] * 20,  # complex,
             "f8": [np.nan, np.inf, -np.inf, 0, np.nan] * 20,  # with infinity
+            "f9": [np.nan] * 100,
         }
     )
 
@@ -2548,6 +2549,11 @@ def test_pandas_profile_summary_stats(dxf: DataExplorerFixture):
             8,
             {"min_value": "-INF", "max_value": "INF"},
         ),
+        (
+            "df1",
+            9,
+            {},
+        ),
         (
             "df_mixed_tz1",
             0,
@@ -3381,6 +3387,7 @@ def test_polars_profile_summary_stats(dxf: DataExplorerFixture):
                 dtype=pl.Datetime("ms", time_zone="UTC"),
             ),  # datetime single tz
             "f7": [np.nan, np.inf, -np.inf, 0, np.nan] * 20,  # with infinity
+            "f8": pl.Series([None] * 100, dtype=pl.Float64),
         }
     )
 
@@ -3468,6 +3475,11 @@ def test_polars_profile_summary_stats(dxf: DataExplorerFixture):
             7,
             {"min_value": "-INF", "max_value": "INF"},
         ),
+        (
+            "df1",
+            8,
+            {},
+        ),
     ]
 
     for table_name, col_index, ex_result in cases: