我有下表:
tableA
+-----------+--------+
| tableA_id | code |
+-----------+--------+
| 1 | code A |
| 2 | code B |
| 3 | code A |
| 3 | code C |
| 3 | code B |
| 4 | code A |
| 4 | code C |
| 4 | code B |
| 5 | code A |
| 5 | code C |
| 5 | code B |
+-----------+--------+
我想使用查询将代码 A、代码 B、代码 C 显示为列标题,然后值将显示 tableA_id 条目在代码字段中是否包含该代码。所以像这样:
+-----------+------------------------------+
| tableA_id | code A | code B | code C |
+-----------+------------------------------+
| 1 | yes | | |
| 2 | | yes | yes |
| 3 | yes | yes | yes |
etc...
你能用 SQL 做到这一点吗?
这个问题的挑战是
code
列可以包含任意值列表。通常 PIVOT 要求将 IN
内的值作为常量提供,尽管使用 Snowflake 可以使用 ANY
或子查询。
这种模式称为 “动态 PIVOT”:
SELECT *
FROM tableA
PIVOT (MIN(code) FOR code IN (ANY)) AS pvt
ORDER BY tableA_id;
带有子查询的版本:
SELECT *
FROM tableA
PIVOT (MIN(code) FOR code IN (SELECT code FROM tableA)) AS pvt
ORDER BY tableA_id;
以前的版本:
使用条件聚合(不同方言之间可移植):
SELECT tableA_id,
MAX(CASE WHEN code ='code A' THEN 'yes' END) AS "code A",
MAX(CASE WHEN code ='code B' THEN 'yes' END) AS "code B",
MAX(CASE WHEN code ='code C' THEN 'yes' END) AS "code C"
FROM tableA
GROUP BY tableA_id;
输出:
╔════════════╦═════════╦═════════╦════════╗
║ tableA_id ║ code A ║ code B ║ code C ║
╠════════════╬═════════╬═════════╬════════╣
║ 1 ║ yes ║ (null) ║ (null) ║
║ 2 ║ (null) ║ yes ║ (null) ║
║ 3 ║ yes ║ yes ║ yes ║
║ 4 ║ yes ║ yes ║ yes ║
║ 5 ║ yes ║ yes ║ yes ║
╚════════════╩═════════╩═════════╩════════╝
有很多可能性(搜索):
PIVOT -> SQL Server/Oracle
CROSSTAB -> Postgresql
SELF OUTER JOIN -> All
CONDITIONAL AGG -> All
...
您需要在 SQL 语法中使用类似的内容
Select *
From Table1
pivot ( Aggregate function For Column Name in ([CodeA], [CodeB] , [CodeC])) as PivotTable
或者如果您想要动态 SQL 语句,您可以尝试以下操作。 您只需更改表和/或列的名称。
declare
cursor code is select code, rownum rn, count(*) over (partition by 1 ) as cnt from (select distinct code from test) order by 1;
build_Case clob;
a varchar2(100);
begin
for i in code loop
if i.rn <> i.cnt then
build_case := build_case || CHR(10) || ' max((case when code = ''' || i.code || ''' then ''yes'' else null end)) ' || i.code || ',';
else
build_case := build_case || CHR(10) || ' max((case when code = ''' || i.code || ''' then ''yes'' else null end)) ' || i.code;
end if;
end loop;
build_case := 'select id,' || build_case || ' from test group by id';
dbms_output.put_line(build_Case);
--execute immediate build_Case;
end;
/