使用 unpivot 从远程 table 插入

Question

因此，在我们的组织中有一个 table 用于安排时间并从中报告数据。日程安排行在一周中的每一天都有一列，其值是 'Y' 或 'N'，具体取决于当天是否有会议。我设法使用 UNPIVOT 以简单的方式完成此操作 select，但不幸的是，我们的需求比这更复杂。

我需要插入一个本地 table 并从远程 table 上的调度 table 查询 DB_Link，这是不可透视的，比如:

SELECT *
FROM   (SELECT class_nbr,
               TERM,
               MON,
               TUES,
               WED,
               THURS,
               FRI,
               SAT,
               SUN
         FROM  schedule_tbl) sched 
UNPIVOT (wkDay FOR DayWeek IN (MON, TUES, WED, THURS, FRI, SAT, SUN)) piv
WHERE   piv.WkDay = 'Y'

同样，当我运行它在 Sql 开发人员中针对 schedule_tbl 所在的服务器时，这工作正常。我得到类似

的结果

Class_nbr TERM DayWeek
1234      Fall MON
1234      Fall Wed
....

然而，令人惊讶的是，当我尝试从

等不同的服务器插入结果时

INSERT INTO other_table (class_nbr, TERM, DayWeek)
SELECT *
FROM   (SELECT class_nbr,
               TERM,
               MON,
               TUES,
               WED,
               THURS,
               FRI,
               SAT,
               SUN
         FROM  schedule_tbl@db_link) sched 
UNPIVOT (wkDay FOR DayWeek IN (MON, TUES, WED, THURS, FRI, SAT, SUN)) piv
WHERE   piv.WkDay = 'Y'

发生的情况是仅插入 DayWeek = 'MON' 的行。应插入的实际星期一行数是。我不确定这是在哪里崩溃的。如果我更改 UNPIVOT 列的顺序，使 TUES 成为第一个，则仅插入星期二的行（同样，正确的数字）。

编辑：我想我可以在这里分享我的实际代码。请记住，如果我将这个确切的查询带到远程数据库并删除 dblink，我得到的不仅仅是星期一的行：

SELECT  '%processinstance',
    0 ,
    crse_id ,
    crse_offer_nbr ,
    strm ,
    session_code ,
    class_section ,
    subject ,
    institution ,
    facility_id ,
    ' ' ,
    ' ' ,
    '0001484' ,
    '4' ,
    PDB_DE0120_TRM_ID ,
    subject || catalog_nbr ,
    strm || class_nbr ,
    to_char ( start_dt , 'YYYYMMDD' ) ,
    to_char ( end_dt , 'YYYYMMDD' ) ,
    DECODE(DAYWEEK,'MON', 'M', 'TUES','T', 'WED','W', 'THURS','R', 'FRI','F', 'SAT','S', 'SUN','X'),
    nvl ( to_char ( meeting_time_start ,'HH24MI' ) ,'9999' ) ,
    nvl ( to_char ( meeting_time_end ,'HH24MI' ) ,'9999' ) ,
    ' ' ,
    ' ' ,
    ' ' ,
    0 ,
    'S' ,
    'C' ,
    decode ( instruction_mode ,'HB' ,'X' ,'DL' ,'X' ,'N' ) ,
    ' ' ,
    ' ' ,
    ' ' ,
    ' ' ,
    ' ' ,
    'E' ,
    SYSDATE ,
    ' ' ,
    SYSDATE,
    ' ' ,
    ' '
FROM    (SELECT A.crse_id ,
                A.crse_offer_nbr ,
                A.strm ,
                A.session_code ,
                A.class_section ,
                B.institution ,
                A.facility_id ,
                B.instruction_mode,
                B.SUBJECT,
                B.CATALOG_NBR,
                A.meeting_time_start,
                A.meeting_time_end,
                B.START_DT,
                B.END_DT,
                B.CLASS_NBR,
                C.PDB_DE0120_TRM_ID,
                MON, 
                TUES, 
                WED, 
                THURS,
                FRI,
                SAT,
                SUN
        FROM    PS_CLASS_MTG_PAT@CSDV1 A
                INNER JOIN  PS_CLASS_TBL@CSDV1 B
                ON          A.STRM = B.STRM AND
                            A.CRSE_ID = B.CRSE_ID AND
                            A.CLASS_SECTION = B.CLASS_SECTION
                INNER JOIN  PS_FSC_SRPT_TRM_VW@CSDV1 C
                ON          C.STRM = B.STRM AND
                            C.ACAD_CAREER = 'CRED'
        WHERE   A.STRM = '2182' AND
                (B.CLASS_STAT <> 'X' OR
                (B.ENRL_TOT > 0 AND
                B.ENRL_STAT <> 'C')) AND
                B.acad_group NOT IN ( '11','12','13','23','80','99' )) B
UNPIVOT EXCLUDE NULLS (wkDay FOR DayWeek IN ( MON AS 'MON', TUES AS 'TUES', WED AS 'WED', THURS AS 'THURS', FRI AS 'FRI', SAT AS 'SAT', SUN AS 'SUN')) piv
WHERE   piv.wkDay = 'Y';

第二次更新：根据要求，我粘贴更正后的查询：

SELECT  '%processinstance',
    0 ,
    crse_id ,
    crse_offer_nbr ,
    strm ,
    session_code ,
    class_section ,
    subject ,
    institution ,
    facility_id ,
    ' ' ,
    ' ' ,
    '0001484' ,
    '4' ,
    PDB_DE0120_TRM_ID ,
    subject || catalog_nbr ,
    strm || class_nbr ,
    to_char ( start_dt , 'YYYYMMDD' ) ,
    to_char ( end_dt , 'YYYYMMDD' ) ,
    DECODE(DAYWEEK,'MON', 'M', 'TUES','T', 'WED','W', 'THURS','R', 'FRI','F', 'SAT','S', 'SUN','X'),
    nvl ( to_char ( meeting_time_start ,'HH24MI' ) ,'9999' ) ,
    nvl ( to_char ( meeting_time_end ,'HH24MI' ) ,'9999' ) ,
    ' ' ,
    ' ' ,
    ' ' ,
    0 ,
    'S' ,
    'C' ,
    decode ( instruction_mode ,'HB' ,'X' ,'DL' ,'X' ,'N' ) ,
    ' ' ,
    ' ' ,
    ' ' ,
    ' ' ,
    ' ' ,
    'E' ,
    ' ' ,
    ' ' ,
    ' '
FROM    (SELECT A.crse_id ,
                A.crse_offer_nbr ,
                A.strm ,
                A.session_code ,
                A.class_section ,
                B.institution ,
                A.facility_id ,
                B.instruction_mode,
                B.SUBJECT,
                B.CATALOG_NBR,
                A.meeting_time_start,
                A.meeting_time_end,
                B.START_DT,
                B.END_DT,
                B.CLASS_NBR,
                C.PDB_DE0120_TRM_ID,
                MON, 
                TUES, 
                WED, 
                THURS,
                FRI,
                SAT,
                SUN
        FROM    PS_CLASS_MTG_PAT@CSDV1 A
                INNER JOIN  PS_CLASS_TBL@CSDV1 B
                ON          A.STRM = B.STRM AND
                            A.CRSE_ID = B.CRSE_ID AND
                            A.CLASS_SECTION = B.CLASS_SECTION
                INNER JOIN  PS_FSC_SRPT_TRM_VW@CSDV1 C
                ON          C.STRM = B.STRM AND
                            C.ACAD_CAREER = 'CRED'
        WHERE   A.STRM = '2182' AND
                (B.CLASS_STAT <> 'X' OR
                (B.ENRL_TOT > 0 AND
                B.ENRL_STAT <> 'C')) AND
                B.acad_group NOT IN ( '11','12','13','23','80','99' )) B
UNPIVOT EXCLUDE NULLS (wkDay FOR DayWeek IN ( MON AS 'MON', TUES AS 'TUES', WED AS 'WED', THURS AS 'THURS', FRI AS 'FRI', SAT AS 'SAT', SUN AS 'SUN')) piv
WHERE   piv.wkDay = 'Y';

Answer 1

这不是一个完全令人满意的解决方案，但它对我有用。如果我从我的 select 子句中删除两个表达式（any 两列），这个问题就会消失。这显然是一个以前未知的涉及 DB Links 的 Oracle 问题，就好像我在表所在的远程服务器上进行了相同的查询并且运行它不存在，并且 Oracle 文档中没有任何关于 DB Links 的内容对这个。我会在某个时候尝试与他们一起提出。

使用 unpivot 从远程 table 插入

Insert from remote table using unpivot

oracle11g

unpivot

dml

dblink