chore(date_trunc): turn panic into unimplemented error for ambiguous time

datafusion/physical-expr/src/datetime_expressions.rs

@@ -581,7 +581,7 @@ fn _date_trunc_coarse_without_tz(
 /// account that some granularities are not uniform durations of time
 /// (e.g. months are not always the same lengths, leap seconds, etc)
 fn date_trunc_coarse(granularity: &str, value: i64, tz: Option<Tz>) -> Result<i64> {
-    let value = match tz {
+    let value: Option<i64> = match tz {
         Some(tz) => {
             // Use chrono DateTime<Tz> to clear the various fields because need to clear per timezone,
             // and NaiveDateTime (ISO 8601) has no concept of timezones
@@ -600,6 +600,14 @@ fn date_trunc_coarse(granularity: &str, value: i64, tz: Option<Tz>) -> Result<i6
         }
     }?;
 
+    // this case is caused by ambiguous time between "daylight saving time" and "standard time" in chrono
+    // https://github.com/apache/arrow-datafusion/issues/8899
+    if tz.is_some() && value.is_none() {
+        return not_impl_err!(
+        "date_trunc that converts ambiguous time between \"daylight saving time\" and \"standard (UTC) time\" is not yet implemented. Suggest to use UTC timezone in date_trunc."
+    );
+    }
+
     // `with_x(0)` are infallible because `0` are always a valid
     Ok(value.unwrap())
 }
@@ -1685,6 +1693,21 @@ mod tests {
                     "2020-09-08T00:00:00+08",
                 ],
             ),
+            (
+                vec![
+                    "2023-10-28T23:00:00+00:00",
+                    "2023-10-29T02:00:00+00:00",
+                    "2024-10-26T23:00:00+00:00",
+                    "2024-10-27T02:00:00+00:00",
+                ],
+                Some("Europe/Berlin".into()),
+                vec![
+                    "2023-10-29T00:00:00+02:00",
+                    "2023-10-29T00:00:00+02:00",
+                    "2024-10-27T00:00:00+02:00",
+                    "2024-10-27T00:00:00+02:00",
+                ],
+            ),
         ];
 
         cases.iter().for_each(|(original, tz_opt, expected)| {
@@ -1716,6 +1739,48 @@ mod tests {
         });
     }
 
+    #[test]
+    fn test_date_trunc_timezones_with_error() {
+        let cases = vec![
+            // daylight saving time ends at 2023-10-29T01:00:00+00:00 in Europe/Berlin
+            (vec!["2023-10-29T00:00:00+00:00"], "second"),
+            (vec!["2023-10-29T00:00:00+00:00"], "minute"),
+            (vec!["2023-10-29T00:00:00+00:00"], "hour"),
+            (vec!["2023-10-29T00:00:00+00:00"], "day"),
+            (vec!["2023-10-29T00:00:00+00:00"], "week"),
+            (vec!["2023-10-29T00:00:00+00:00"], "month"),
+            (vec!["2023-10-29T00:00:00+00:00"], "quarter"),
+            (vec!["2023-10-29T00:00:00+00:00"], "year"),
+            // daylight saving time ends at 2024-10-27T01:00:00+00:00 in Europe/Berlin
+            (vec!["2024-10-27T00:00:00+00:00"], "second"),
+            (vec!["2024-10-27T00:00:00+00:00"], "minute"),
+            (vec!["2024-10-27T00:00:00+00:00"], "hour"),
+            (vec!["2024-10-27T00:00:00+00:00"], "day"),
+            (vec!["2024-10-27T00:00:00+00:00"], "week"),
+            (vec!["2024-10-27T00:00:00+00:00"], "month"),
+            (vec!["2024-10-27T00:00:00+00:00"], "quarter"),
+            (vec!["2024-10-27T00:00:00+00:00"], "year"),
+        ];
+
+        cases.iter().for_each(|(original, granularity)| {
+            let input = original
+                .iter()
+                .map(|s| Some(string_to_timestamp_nanos(s).unwrap()))
+                .collect::<TimestampNanosecondArray>()
+                .with_timezone_opt("Europe/Berlin".into());
+            let result = date_trunc(&[
+                ColumnarValue::Scalar(ScalarValue::from(*granularity)),
+                ColumnarValue::Array(Arc::new(input)),
+            ]);
+            assert!(result.is_err());
+
+            assert!(matches!(
+                result.unwrap_err(),
+                DataFusionError::NotImplemented(_)
+            ));
+        });
+    }
+
     #[test]
     fn test_date_bin_single() {
         use chrono::Duration;