fix comm

Author: xxhZs

e2e_test/batch/types/timestamp_ns.slt.part

@@ -5,7 +5,7 @@ statement ok
 create table t1(v1 int, v2 timestamp);
 
 statement ok
-insert into t1 values(1,'2013-01-01 01:01:01.123456789'),(2,'2012-01-01 01:01:01.123456'),(3,'0000-01-01 01:01:01.123456789'),(4,'2213-01-01 01:01:01.123456789'),(5,null);
+insert into t1 values(1,'2013-01-01 01:01:01.123456789'),(2,'2012-01-01 01:01:01.123456'),(3,'0000-01-01 01:01:01.123456789'),(4,'2213-01-01 01:01:01.123456789'),(5,null),(6,'2013-01-01 01:01:01.123456789');
 
 query T rowsort
 select * from t1;
@@ -15,6 +15,7 @@ select * from t1;
 3  0001-01-01 01:01:01.123456789 BC
 4  2213-01-01 01:01:01.123456789
 5  NULL
+6  2013-01-01 01:01:01.123456789
 
 query T
 select * from t1 where v2 is null;
@@ -35,6 +36,7 @@ from t1;
 3  0001-01-01 01:01:01.123456789 BC  Before 2000
 4  2213-01-01 01:01:01.123456789     Future
 5  NULL                              Future
+6  2013-01-01 01:01:01.123456789     21st Century
 
 query T rowsort
 select v1, v2, coalesce(v2, '1900-01-01 00:00:00') as coalesce_v2 from t1;
@@ -44,6 +46,7 @@ select v1, v2, coalesce(v2, '1900-01-01 00:00:00') as coalesce_v2 from t1;
 3  0001-01-01 01:01:01.123456789 BC  0001-01-01 01:01:01.123456789 BC
 4  2213-01-01 01:01:01.123456789     2213-01-01 01:01:01.123456789
 5  NULL                              1900-01-01 00:00:00
+6  2013-01-01 01:01:01.123456789     2013-01-01 01:01:01.123456789
 
 query T
 select count(v2) as total_rows from t1;
@@ -58,13 +61,15 @@ select * from t1 order by v2;
 3  0001-01-01 01:01:01.123456789 BC
 4  2213-01-01 01:01:01.123456789
 5  NULL
+6  2013-01-01 01:01:01.123456789
 
 query T rowsort
 select * from t1 where v2 >= '2012-01-01 01:01:01.123456';
 ----
 1  2013-01-01 01:01:01.123456789
 2  2012-01-01 01:01:01.123456
 4  2213-01-01 01:01:01.123456789
+6  2013-01-01 01:01:01.123456789
 
 query T rowsort
 select v1, cast(v2 as date) as date_v2, cast(v2 as timestamp with time zone) as timestamptz_v2 from t1;
@@ -74,6 +79,7 @@ select v1, cast(v2 as date) as date_v2, cast(v2 as timestamp with time zone) as
 3  0001-01-01 BC  0001-01-01 01:01:01.123456+00:00 BC
 4  2213-01-01     2213-01-01 01:01:01.123456+00:00
 5  NULL           NULL
+6  2013-01-01     2013-01-01 01:01:01.123456+00:00
 
 query T rowsort
 select v1, date_trunc('day', v2) AS truncated_v2 from t1;
@@ -83,6 +89,7 @@ select v1, date_trunc('day', v2) AS truncated_v2 from t1;
 3  0001-01-01 00:00:00 BC
 4  2213-01-01 00:00:00
 5  NULL
+6  2013-01-01 00:00:00
 
 query T rowsort
 select v1, v2 at time zone 'UTC' as v2_utc from t1;
@@ -92,6 +99,7 @@ select v1, v2 at time zone 'UTC' as v2_utc from t1;
 3  0001-01-01 01:01:01.123456+00:00 BC
 4  2213-01-01 01:01:01.123456+00:00
 5  NULL
+6  2013-01-01 01:01:01.123456+00:00
 
 query T rowsort
 select v1, to_char(v2, 'YYYY-MM-DD HH24:MI:SS.NS') as formatted_v2 from t1;
@@ -101,6 +109,7 @@ select v1, to_char(v2, 'YYYY-MM-DD HH24:MI:SS.NS') as formatted_v2 from t1;
 3  0000-01-01 01:01:01.123456789
 4  2213-01-01 01:01:01.123456789
 5  NULL
+6  2013-01-01 01:01:01.123456789
 
 query T rowsort
 select generate_series('2013-01-01 01:01:01.123456789'::timestamp,'2013-01-01 01:01:05.123456790'::timestamp, '1 s');
@@ -111,5 +120,93 @@ select generate_series('2013-01-01 01:01:01.123456789'::timestamp,'2013-01-01 01
 2013-01-01 01:01:04.123456789
 2013-01-01 01:01:05.123456789
 
+query T rowsort
+select DISTINCT v2 FROM t1;
+----
+0001-01-01 01:01:01.123456789 BC
+2012-01-01 01:01:01.123456
+2013-01-01 01:01:01.123456789
+2213-01-01 01:01:01.123456789
+NULL
+
+query T rowsort
+select v2, count(*) from t1 group by v2;
+----
+0001-01-01 01:01:01.123456789 BC  1
+2012-01-01 01:01:01.123456        1
+2013-01-01 01:01:01.123456789     2
+2213-01-01 01:01:01.123456789     1
+null                              1
+
+query T rowsort
+select * from t1 order by v2 desc;
+----
+5  NULL
+4  2213-01-01 01:01:01.123456789
+6  2013-01-01 01:01:01.123456789
+1  2013-01-01 01:01:01.123456789
+2  2012-01-01 01:01:01.123456
+3  0001-01-01 01:01:01.123456789 BC
+
+query T rowsort
+select max(v2) from t1;
+----
+2213-01-01 01:01:01.123456789
+
+query T rowsort
+select v1, extract(epoch from v2) from t1;
+----
+1  1357002061.123456789
+2  1325379661.123456000
+3  -62167215538.876544
+4  7668349261.123456789
+5  NULL
+6  1357002061.123456789
+
+query T rowsort
+select v1, extract(second from v2) from t1;
+----
+1  1.123456789
+2  1.123456000
+3  1.123456789
+4  1.123456789
+5  NULL
+6  1.123456789
+
+query T rowsort
+select v1, extract(millisecond from v2) from t1;
+----
+1  1123.456789
+2  1123.456000
+3  1123.456789
+4  1123.456789
+5  NULL
+6  1123.456789
+
+query T rowsort
+select v1, extract(microsecond from v2) from t1;
+----
+1  1123456.789
+2  1123456.000
+3  1123456.789
+4  1123456.789
+5  NULL
+6  1123456.789
+
+query T rowsort
+select v1, extract(nanosecond from v2) from t1;
+----
+1  1123456789
+2  1123456000
+3  1123456789
+4  1123456789
+5  NULL
+6  1123456789
+
+query T rowsort
+select make_timestamp(2013, 01, 01, 01, 01, 1.123456789);
+----
+2013-01-01 01:01:01.123456789
+
 statement ok
 drop table t1;

src/common/src/types/datetime.rs

@@ -256,7 +256,7 @@ enum ErrorKind {
     ParseDate,
     #[error("Can't cast string to time (expected format is HH:MM:SS[.D+{{up to 6 digits}}][Z] or HH:MM)")]
     ParseTime,
-    #[error("Can't cast string to timestamp (expected format is YYYY-MM-DD HH:MM:SS[.D+{{up to 6 digits}}] or YYYY-MM-DD HH:MM or YYYY-MM-DD or ISO 8601 format)")]
+    #[error("Can't cast string to timestamp (expected format is YYYY-MM-DD HH:MM:SS[.D+{{up to 9 digits}}] or YYYY-MM-DD HH:MM or YYYY-MM-DD or ISO 8601 format)")]
     ParseTimestamp,
 }
 
@@ -486,6 +486,7 @@ impl Time {
 //   it is *microseconds* since 1970-01-01 midnight. 2^62 microseconds covers 146235 years.
 // * When the highest 2 bits are `10` or `01`, we flip the second bit to get values from `0b1100...00` to `0b0011..11` again.
 //   It is *seconds* since 1970-01-01 midnight. It is then followed by another 32 bits as nanoseconds within a second.
+// Since timestamp is negative when it is less than 1970-1-1, you need to take both cases into account(`11+00`` or `01+10``).
 enum FirstI64 {
     V0 { usecs: i64 },
     V1 { secs: i64 },

src/expr/impl/src/scalar/extract.rs

@@ -42,17 +42,18 @@ fn extract_from_datelike(date: impl Datelike, unit: Unit) -> Decimal {
 
 /// Extract field from `Timelike`.
 fn extract_from_timelike(time: impl Timelike, unit: Unit) -> Decimal {
-    let usecs = || time.second() as u64 * 1_000_000 + (time.nanosecond() / 1000) as u64;
+    let usecs = || time.second() as u64 * 1_000_000_000 + time.nanosecond() as u64;
     match unit {
         Hour => time.hour().into(),
         Minute => time.minute().into(),
-        Second => Decimal::from_i128_with_scale(usecs() as i128, 6),
-        Millisecond => Decimal::from_i128_with_scale(usecs() as i128, 3),
-        Microsecond => usecs().into(),
+        Second => Decimal::from_i128_with_scale(usecs() as i128, 9),
+        Millisecond => Decimal::from_i128_with_scale(usecs() as i128, 6),
+        Microsecond => Decimal::from_i128_with_scale(usecs() as i128, 3),
+        Nanosecond => usecs().into(),
         Epoch => {
-            let usecs = time.num_seconds_from_midnight() as u64 * 1_000_000
-                + (time.nanosecond() / 1000) as u64;
-            Decimal::from_i128_with_scale(usecs as i128, 6)
+            let usecs =
+                time.num_seconds_from_midnight() as u64 * 1_000_000_000 + time.nanosecond() as u64;
+            Decimal::from_i128_with_scale(usecs as i128, 9)
         }
         u => unreachable!("invalid unit {:?} for time", u),
     }
@@ -92,12 +93,20 @@ fn extract_from_time(time: Time, unit: &Unit) -> Decimal {
 fn extract_from_timestamp(timestamp: Timestamp, unit: &Unit) -> Decimal {
     match unit {
         Epoch => {
-            let epoch = timestamp.0.and_utc().timestamp_micros();
-            Decimal::from_i128_with_scale(epoch as i128, 6)
+            if let Some(nanos) = timestamp.0.and_utc().timestamp_nanos_opt() {
+                Decimal::from_i128_with_scale(nanos as i128, 9)
+            } else {
+                let micros = timestamp.0.and_utc().timestamp_micros();
+                Decimal::from_i128_with_scale(micros as i128, 6)
+            }
         }
         Julian => {
-            let epoch =
-                Decimal::from_i128_with_scale(timestamp.0.and_utc().timestamp_micros() as i128, 6);
+            let epoch = if let Some(nanos) = timestamp.0.and_utc().timestamp_nanos_opt() {
+                Decimal::from_i128_with_scale(nanos as i128, 9)
+            } else {
+                let epoch = timestamp.0.and_utc().timestamp_micros();
+                Decimal::from_i128_with_scale(epoch as i128, 6)
+            };
             epoch / (24 * 60 * 60).into() + 2_440_588.into()
         }
         _ if unit.is_date_unit() => extract_from_datelike(timestamp.0.date(), *unit),
@@ -284,6 +293,7 @@ define_unit! {
         Second,
         Millisecond,
         Microsecond,
+        Nanosecond,
         Epoch,
         Julian,
         Timezone,
@@ -307,7 +317,7 @@ impl Unit {
     const fn is_time_unit(self) -> bool {
         matches!(
             self,
-            Hour | Minute | Second | Millisecond | Microsecond | Epoch
+            Hour | Minute | Second | Millisecond | Microsecond | Nanosecond | Epoch
         )
     }
 

src/expr/impl/src/scalar/make_time.rs

@@ -43,8 +43,8 @@ fn make_naive_time(hour: i32, min: i32, sec: F64) -> Result<NaiveTime> {
         });
     }
     let sec_u32 = sec.0.trunc() as u32;
-    let microsecond_u32 = ((sec.0 - sec.0.trunc()) * 1_000_000.0).round_ties_even() as u32;
-    NaiveTime::from_hms_micro_opt(hour as u32, min as u32, sec_u32, microsecond_u32).ok_or_else(
+    let nanosecond_u32 = ((sec.0 - sec.0.trunc()) * 1_000_000_000.0).round_ties_even() as u32;
+    NaiveTime::from_hms_nano_opt(hour as u32, min as u32, sec_u32, nanosecond_u32).ok_or_else(
         || ExprError::InvalidParam {
             name: "hour, min, sec",
             reason: format!("invalid time: {}:{}:{}", hour, min, sec).into(),