在 SQL 中使用 PIVOT 和 UNPIVOT 进行数据重组

Question

我有一张表，其中的数据与原始数据相似。 通过使用原始数据，我想在 SQL 中使用 PIVOT 或 UNPIVOT 重组为修改后的数据。

原始数据

lot	table1	table2	table3	table4	...	table28	headno	…
a	xxa	xxb	xxc	xxd	...	xxf	1	…
b	xxg	xxe	xxi	xxp	...	xxx	1	…
c	xxk	xxm	xxr	xxt	...	xxu	4	…
…	...	...	...	...	...	...	…	…

修改后的结果

lot	table	table NO	head no	…
a	xxa	table1	1	…
a	xxb	table2	1	…
a	xxc	table3	1	…
a	xxd	table4	1	…
a	...	….	…	…
b	xxg	table1	1	…
b	xxe	table2	1	…
b	xxi	table3	1	…
b	xxp	table4	1	…
b	...	…	…	…
c	xxk	table1	4	…
…	…	…	…	…

提前致谢。

Original data and Modified data

Original data and Modified data Updated version

Answer 1

我的反透视技术是遵守通用 SQL 标准（不限于在特定数据库上运行的标准）。

在这种情况下，取消透视是关于将“水平”表与内联表 i 交叉连接，该表由整数列 i 和与水平列一样多的连续整数组成根据当前结果集中 i 的当前值，透视 - 和 CASE WHEN 表达式以使用右列：

WITH
-- your input - 5 columns work the same as 28 ...
-- do not use in your real query.
indata(lot,table1,table2,table3,table4,table5,headno) AS (
          SELECT 'a','xxa','xxb','xxc','xxd','xxf',1 FROM dual
UNION ALL SELECT 'b','xxg','xxe','xxi','xxp','xxx',1 FROM dual
UNION ALL SELECT 'c','xxk','xxm','xxr','xxt','xxu',4 FROM dual
)
-- real query starts here, replace following comma with "WITH"
,
-- need series of integers - 28 rather than 5 in the full example ..
i(i) AS (
          SELECT 1  FROM dual
UNION ALL SELECT 2  FROM dual
UNION ALL SELECT 3  FROM dual
UNION ALL SELECT 4  FROM dual
UNION ALL SELECT 5  FROM dual
)
SELECT
  lot 
, CASE i 
    WHEN 1 THEN table1 
    WHEN 2 THEN table2 
    WHEN 3 THEN table3 
    WHEN 4 THEN table4 
    WHEN 5 THEN table5 
    ELSE NULL
   END AS tbl 
,  'table'||CAST(i AS VARCHAR(1)) AS table_NO
, headno
FROM indata CROSS JOIN i 
ORDER BY lot,i
-- out  lot | tbl | table_NO | headno 
-- out -----+-----+----------+--------
-- out  a   | xxa | table1   |      1
-- out  a   | xxb | table2   |      1
-- out  a   | xxc | table3   |      1
-- out  a   | xxd | table4   |      1
-- out  a   | xxf | table5   |      1
-- out  b   | xxg | table1   |      1
-- out  b   | xxe | table2   |      1
-- out  b   | xxi | table3   |      1
-- out  b   | xxp | table4   |      1
-- out  b   | xxx | table5   |      1
-- out  c   | xxk | table1   |      4
-- out  c   | xxm | table2   |      4
-- out  c   | xxr | table3   |      4
-- out  c   | xxt | table4   |      4
-- out  c   | xxu | table5   |      4

Answer 2

你需要UNPIVOT:

SELECT lot,
       "TABLE",
       tableno,
       headno
       /*,...*/
FROM   table_name
UNPIVOT (
  "TABLE" FOR tableno IN (
    table1, table2, table3, table4, /*...,*/ table28
  )
)

其中，对于样本数据：

CREATE TABLE table_name (lot, table1, table2, table3, table4, /*...,*/ table28, headno /*,...*/ ) AS
SELECT 'a', 'xxa', 'xxb', 'xxc', 'xxd', /*...,*/ 'xxf', 1 /*,...*/ FROM DUAL UNION ALL
SELECT 'b', 'xxg', 'xxe', 'xxi', 'xxp', /*...,*/ 'xxx', 1 /*,...*/ FROM DUAL UNION ALL
SELECT 'c', 'xxk', 'xxm', 'xxr', 'xxt', /*...,*/ 'xxu', 4 /*,...*/ FROM DUAL;

输出：

LOT	TABLE	TABLENO	HEADNO
a	xxa	TABLE1	1
a	xxb	TABLE2	1
a	xxc	TABLE3	1
a	xxd	TABLE4	1
a	xxf	TABLE28	1
b	xxg	TABLE1	1
b	xxe	TABLE2	1
b	xxi	TABLE3	1
b	xxp	TABLE4	1
b	xxx	TABLE28	1
c	xxk	TABLE1	4
c	xxm	TABLE2	4
c	xxr	TABLE3	4
c	xxt	TABLE4	4
c	xxu	TABLE28	4

db小提琴here