按纪元时间分组

Question

我需要按小时分组以获得转换后的纪元时间，这是我所做的，但它仍然是按天分组。有人可以支持吗？

 select count (*) as "Total",to_char(trunc(to_date('19700101 00', 'YYYYMMDD HH24') + ( 1 / 24 / 60 / 60 ) * INSERTION_TIME),'DD/MM/YYYY HH24') as "Day"
from Records
where ID=35 and (INSERTION_TIME/86400)+ to_date('01-01-1970 00:00:00','dd-mm-yyyy HH24:MI:SS') >=sysdate - 4
group by trunc(to_date('19700101 00', 'YYYYMMDD HH24') + ( 1 / 24 / 60 / 60 ) * INSERTION_TIME)
order by trunc(to_date('19700101 00', 'YYYYMMDD HH24') + ( 1 / 24 / 60 / 60 ) * INSERTION_TIME)

Answer 1

注意 trunc 在你的表情上的行为。它删除了小时部分。

测试用例 dbfiddle.uk:

Test case

未分组的 SQL，在 select 列表中有额外的表达式：

select to_char(trunc(to_date('19700101 00', 'YYYYMMDD HH24') + ( 1 / 24 / 60 / 60 ) * INSERTION_TIME),'DD/MM/YYYY HH24') as "Day"
     , to_char(     (to_date('19700101 00', 'YYYYMMDD HH24') + ( 1 / 24 / 60 / 60 ) * INSERTION_TIME),'DD/MM/YYYY HH24') as "Day"
  from Records
 where ID = 35 and (INSERTION_TIME/86400) + to_date('01-01-1970 00:00:00','dd-mm-yyyy HH24:MI:SS') >= sysdate - 4
   AND rownum < 6
 order by      (to_date('19700101 00', 'YYYYMMDD HH24') + ( 1 / 24 / 60 / 60 ) * INSERTION_TIME)
;

未分组的结果：

现在添加GROUP BY:

select COUNT(*) AS n
     , to_char(trunc(to_date('19700101 00', 'YYYYMMDD HH24') + ( 1 / 24 / 60 / 60 ) * INSERTION_TIME),'DD/MM/YYYY HH24') as "Day"
     , to_char(     (to_date('19700101 00', 'YYYYMMDD HH24') + ( 1 / 24 / 60 / 60 ) * INSERTION_TIME),'DD/MM/YYYY HH24') as "Day"
  from Records
 where ID = 35 and (INSERTION_TIME/86400) + to_date('01-01-1970 00:00:00','dd-mm-yyyy HH24:MI:SS') >= sysdate - 4
 group by      (to_date('19700101 00', 'YYYYMMDD HH24') + ( 1 / 24 / 60 / 60 ) * INSERTION_TIME)
 order by      (to_date('19700101 00', 'YYYYMMDD HH24') + ( 1 / 24 / 60 / 60 ) * INSERTION_TIME)
;

分组结果：

创建测试用例：

CREATE TABLE Records (
    id              int
  , insertion_time  int
);

INSERT INTO Records VALUES (35, 4011134567);
INSERT INTO Records VALUES (35, 4011134567);
INSERT INTO Records VALUES (35, 4011134567);
INSERT INTO Records VALUES (35, 4011134567);

Answer 2

您可以去掉大部分算术运算，只需使用：

SELECT COUNT(*) as "Total",
       TO_CHAR(
         TRUNC(DATE '1970-01-01' + INTERVAL '1' SECOND * INSERTION_TIME, 'HH'),
         'DD/MM/YYYY HH24'
       ) as "Day"
FROM   Records
WHERE  ID=35
AND    DATE '1970-01-01' + INTERVAL '1' SECOND * INSERTION_TIME >=sysdate - 4
GROUP BY
       TRUNC(DATE '1970-01-01' + INTERVAL '1' SECOND * INSERTION_TIME, 'HH')
ORDER BY
       TRUNC(DATE '1970-01-01' + INTERVAL '1' SECOND * INSERTION_TIME, 'HH')

注意：纪元时间通常为 1970-01-01 00:00:00 UTC。如果您的会话时区不是 UTC，那么您可能会错误地计算纪元时间并希望改用时间戳：

SELECT COUNT(*) as "Total",
       TO_CHAR(
         TRUNC(
           TIMESTAMP '1970-01-01 00:00:00 UTC' + INTERVAL '1' SECOND * INSERTION_TIME,
           'HH'
         ),
         'DD/MM/YYYY HH24'
       ) as "Day"
FROM   Records
WHERE  ID=35
AND    TIMESTAMP '1970-01-01 00:00:00 UTC' + INTERVAL '1' SECOND * INSERTION_TIME
         >=SYSTIMESTAMP - INTERVAL '4' DAY
GROUP BY
       TRUNC(
         TIMESTAMP '1970-01-01 00:00:00 UTC' + INTERVAL '1' SECOND * INSERTION_TIME,
         'HH'
       )
ORDER BY
       TRUNC(
         TIMESTAMP '1970-01-01 00:00:00 UTC' + INTERVAL '1' SECOND * INSERTION_TIME,
         'HH'
       )

按纪元时间分组

Group by Hour for Epoch time

sql

oracle

epoch