DB2 SQL 添加列值

Question

我正在尝试将一些计算列添加在一起，其中来自这些列的值是当前记录和下一条记录之间时间戳差异函数的子字符串。 不幸的是，子字符串确实返回了空值，我试图通过 COALESCE、ISNULL 来转换这些值，但为了正确添加这些值，我没有得到任何帮助。 我也尝试过 SUM 函数，但我也遇到了错误。

select T01."ENGVIDN", T01."ENGSTAT", (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8)))) as "T01 TIMESTAMP",  (TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) as "T02 TIMESTAMP",  
CAST(((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))) as INT) as "DIFF",
CAST(SUBSTR(((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))), LOCATE('.', ((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))))-8, 2)as INT) as "DAYS",
CAST(SUBSTR(((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))), LOCATE('.', ((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))))-6, 2) as INT) as "HOURS",
CAST(SUBSTR(((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))), LOCATE('.', ((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))))-4, 2) as INT) as "MINUTES",
CAST(SUBSTR(((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))), LOCATE('.', ((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))))-2, 2) as INT) as "SECONDS",
--
(CAST(SUBSTR(((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))), LOCATE('.', ((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))))-4, 2) as INT)*60 +
CAST(SUBSTR(((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))), LOCATE('.', ((TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) - (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))))-2, 2) as INT))



from mrc_main."SAMENGST" T01 
        INNER JOIN mrc_main."SAMENGST" T02 on RRN(T02) = RRN(T01)+1
                and (TIMESTAMP(((SUBSTR( T02."ENGSTME", 1, LOCATE('T', T02."ENGSTME")-1))),(SUBSTR( T02."ENGSTME", LOCATE('T', T02."ENGSTME")+1, 8)))) > (TIMESTAMP(((SUBSTR( T01."ENGSTME", 1, LOCATE('T', T01."ENGSTME")-1))),(SUBSTR( T01."ENGSTME", LOCATE('T', T01."ENGSTME")+1, 8))))


WHERE T01."ENGVIDN" = T02."ENGVIDN" and
        T02."ENGVIDN" = 212014918948687 and 
        T01."ENGSTME" like '%2020-02-17%' 

我正在尝试添加标记为 DAYS、HOURS、MINUTES、SECONDS 的列，那么完成此操作的正确方法是什么？

下面的示例数据

ENGVIDN         ENGSTME                 ENGSTME
212014918948687 2020-02-17T09:46:19Z    Off
212014918948687 2020-02-17T09:46:29Z    On
212014918948687 2020-02-17T09:50:10Z    Idle
212014918948687 2020-02-17T10:01:11Z    On
212014918948687 2020-02-17T10:12:16Z    Idle
212014918948687 2020-02-17T11:10:17Z    On
212014918948687 2020-02-17T11:19:27Z    Idle
212014918948687 2020-02-17T11:23:27Z    On
212014918948687 2020-02-17T11:35:19Z    Idle
212014918948687 2020-02-17T11:36:53Z    Off
212014918948687 2020-02-17T12:07:20Z    On
212014918948687 2020-02-17T12:09:20Z    Idle
212014918948687 2020-02-17T12:21:01Z    On
212014918948687 2020-02-17T12:38:48Z    Idle
212014918948687 2020-02-17T12:39:47Z    On
212014918948687 2020-02-17T12:53:44Z    Idle
212014918948687 2020-02-17T12:56:23Z    On
212014918948687 2020-02-17T12:58:48Z    Idle
212014918948687 2020-02-17T13:01:14Z    On
212014918948687 2020-02-17T13:17:17Z    Idle
212014918948687 2020-02-17T14:01:01Z    On
212014918948687 2020-02-17T14:09:16Z    Idle
212014918948687 2020-02-17T14:41:29Z    On
212014918948687 2020-02-17T15:01:19Z    Off
212014918948687 2020-02-17T15:35:42Z    On
212014918948687 2020-02-17T15:43:11Z    Off
212014918948687 2020-02-17T16:16:57Z    On
212014918948687 2020-02-17T16:26:36Z    Idle
212014918948687 2020-02-17T16:50:30Z    On
212014918948687 2020-02-17T16:58:20Z    Idle
212014918948687 2020-02-17T17:03:24Z    On
212014918948687 2020-02-17T17:13:30Z    Idle
212014918948687 2020-02-17T17:23:16Z    On
212014918948687 2020-02-17T17:36:51Z    Off
212014918948687 2020-02-17T19:00:35Z    On
212014918948687 2020-02-17T19:19:31Z    Off
212014918948687 2020-02-17T19:21:58Z    On
212014918948687 2020-02-17T19:24:09Z    Idle
212014918948687 2020-02-17T19:26:08Z    On
212014918948687 2020-02-17T19:42:24Z    Idle
212014918948687 2020-02-17T19:44:27Z    On
212014918948687 2020-02-17T19:48:55Z    Off
212014918948687 2020-02-17T19:50:53Z    On
212014918948687 2020-02-17T19:57:27Z    Off
212014918948687 2020-02-18T01:55:56Z    On
212014918948687 2020-02-18T01:58:43Z    Off
212014918948687 2020-02-18T05:02:17Z    Off
212014918948687 2020-02-18T08:58:01Z    On
212014918948687 2020-02-18T09:01:45Z    Idle
212014918948687 2020-02-18T09:05:17Z    On

Answer 1

我不确定，但似乎是 V7.1。
试试这个：

WITH TAB (ENGSTME) AS 
(
VALUES
  TIMESTAMP('2020-02-17-09.46.19')
, TIMESTAMP('2020-02-17-09.46.29')
, TIMESTAMP('2020-02-17-09.50.10')
, TIMESTAMP('2020-02-17-10.01.11')
, TIMESTAMP('2020-02-17-10.12.16')
, TIMESTAMP('2020-02-17-11.10.17')
)
, TAB_ENUM (RN, ENGSTME) AS 
(
SELECT ROWNUMBER() OVER (ORDER BY ENGSTME) AS RN, ENGSTME
FROM TAB
)
SELECT 
  A.ENGSTME, B.ENGSTME AS ENGSTME_PREV
, DIGITS(DEC(A.ENGSTME - B.ENGSTME, 14)) AS TS_DURATION --yyyymmddhhmmss format with leading zeroes
, (DAYS(A.ENGSTME) - DAYS(B.ENGSTME)) * 86400 + MIDNIGHT_SECONDS(A.ENGSTME) - MIDNIGHT_SECONDS(B.ENGSTME) AS TS_DIFF_SEC
FROM TAB_ENUM A
LEFT JOIN TAB_ENUM B ON B.RN = A.RN - 1 
ORDER BY A.ENGSTME;

结果是：

|ENGSTME            |ENGSTME_PREV       |TS_DURATION   |TS_DIFF_SEC|
|-------------------|-------------------|--------------|-----------|
|2020-02-17 09:46:19|                   |              |           |
|2020-02-17 09:46:29|2020-02-17 09:46:19|00000000000010|10         |
|2020-02-17 09:50:10|2020-02-17 09:46:29|00000000000341|221        |
|2020-02-17 10:01:11|2020-02-17 09:50:10|00000000001101|661        |
|2020-02-17 10:12:16|2020-02-17 10:01:11|00000000001105|665        |
|2020-02-17 11:10:17|2020-02-17 10:12:16|00000000005801|3481       |

这个想法是首先枚举行，然后重新加入行以获取当前行的附加列中的先前时间戳。 您可以对每个组（如果有）进行这样的枚举，而不是像示例中那样对整个数据集进行枚举（在OVER开头添加相应的PARTITION BY子句，并在JOIN中添加此类组的列）。
您可以使用TS_DURATION列或在TS_DIFF_SEC上使用您自己的计算来获得您想要的其他列。

请提供此示例数据所需的确切结果，如果这不是您想要的。