A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.
Python artificial-intelligencedagdata-sciencedata-visualizationdataflowdeveloper-toolsmachine-learningnotebookspipelinepythonreactivesqlweb-app
Fix OSError when displaying Polars DataFrames with old timestamps on Windows (#7477)
Myles Scolnick6280cd9f26ec
diff --git a/marimo/_plugins/ui/_impl/tables/narwhals_table.py b/marimo/_plugins/ui/_impl/tables/narwhals_table.pyindex f4b49f87b0..4c142a8f50 100644--- a/marimo/_plugins/ui/_impl/tables/narwhals_table.py+++ b/marimo/_plugins/ui/_impl/tables/narwhals_table.py@@ -616,9 +616,36 @@ class NarwhalsTableManager(int(hours), int(minutes), int(seconds), int(microseconds))elif dtype == nw.Date:- bin_end = datetime.date.fromtimestamp(bin_end / ms_time)+ # Use timedelta to handle dates before Unix epoch (1970)+ # which cause OSError on Windows with fromtimestamp+ try:+ bin_end = datetime.date.fromtimestamp(bin_end / ms_time)+ except (OSError, OverflowError, ValueError):+ # Fall back to timedelta calculation for old dates+ epoch = datetime.datetime(+ 1970, 1, 1, tzinfo=datetime.timezone.utc+ )+ bin_end_dt = epoch + datetime.timedelta(+ seconds=bin_end / ms_time+ )+ bin_end = bin_end_dt.date()else:- bin_end = datetime.datetime.fromtimestamp(bin_end / ms_time)+ # Use timedelta to handle datetimes before Unix epoch (1970)+ # which cause OSError on Windows with fromtimestamp+ try:+ bin_end = datetime.datetime.fromtimestamp(+ bin_end / ms_time+ )+ except (OSError, OverflowError, ValueError):+ # Fall back to timedelta calculation for old dates+ epoch = datetime.datetime(+ 1970, 1, 1, tzinfo=datetime.timezone.utc+ )+ bin_end = epoch + datetime.timedelta(+ seconds=bin_end / ms_time+ )+ # Remove timezone to match fromtimestamp behavior+ bin_end = bin_end.replace(tzinfo=None)# Only append if the count is greater than 0if count > 0:diff --git a/tests/_plugins/ui/_impl/tables/test_narwhals.py b/tests/_plugins/ui/_impl/tables/test_narwhals.pyindex e69e15bceb..6b3196f879 100644--- a/tests/_plugins/ui/_impl/tables/test_narwhals.py+++ b/tests/_plugins/ui/_impl/tables/test_narwhals.py@@ -1315,6 +1315,57 @@ class TestGetBinValuesTemporal:assert bin_values[0].count == 5[email protected](not HAS_DEPS, reason="optional dependencies not installed")+ "df",+ create_dataframes(+ {+ "old_dates": [+ datetime.datetime.fromisoformat("1902-01-01 00:00:00"),+ datetime.datetime.fromisoformat("2020-01-01 00:00:00"),+ datetime.datetime.fromisoformat("2020-01-02 00:00:00"),+ datetime.datetime.fromisoformat("2020-01-03 00:00:00"),+ datetime.datetime.fromisoformat("2020-01-04 00:00:00"),+ datetime.datetime.fromisoformat("2020-01-05 00:00:00"),+ datetime.datetime.fromisoformat("2020-01-06 00:00:00"),+ datetime.datetime.fromisoformat("2020-01-07 00:00:00"),+ datetime.datetime.fromisoformat("2020-01-08 00:00:00"),+ datetime.datetime.fromisoformat("2020-01-09 00:00:00"),+ datetime.datetime.fromisoformat("2020-01-10 00:00:00"),+ ]+ },+ exclude=["ibis"],+ ),+)+class TestOldDates:+ """Test handling of old dates (pre-1970) that cause OSError on Windows."""++ def test_get_bin_values_with_old_dates(self, df: Any) -> None:+ """+ Test that get_bin_values works with timestamps before Unix epoch.++ This reproduces issue #7469 where dates like 1902-01-01 cause+ OSError: [Errno 22] Invalid argument on Windows when using+ datetime.fromtimestamp().+ """+ manager = NarwhalsTableManager.from_dataframe(df)+ # Should not raise OSError even with old dates+ bin_values = manager.get_bin_values("old_dates", 3)++ # Verify we get valid bin values+ assert len(bin_values) > 0+ assert all(+ isinstance(bv.bin_start, (datetime.datetime, type(None)))+ for bv in bin_values+ )+ assert all(+ isinstance(bv.bin_end, (datetime.datetime, type(None)))+ for bv in bin_values+ )+ # Total count should be 11 (all rows)+ assert sum(bin_value.count for bin_value in bin_values) == 11++@pytest.mark.skipif(not HAS_DEPS, reason="optional dependencies not installed")@pytest.mark.parametrize("df",