Question

下面是

场景：

(* denotes primary key)
**CompanyUnit**(*Unit_id, unit_loc,Year);

**Employees**(Unit_id,Dept_id,no_of_emp); (unit_id and dept_id are foreign keys)

**ref_department**(*dept_id,dept_name,dept_desc);

示例数据：

unit_id       unit_loc        year
------------------------------------
1             Delhi            2003
2             Mumbai           2004
------------------------------------
dept_id       dept_name        dept_desc
----------------------------------------
101           ABC              ABC-AI
102           ABC              ABC-BI
103           ABC              ABC-CS
104           XYZ              XYZ-Testing
105           XYZ              XYZ-Development
----------------------------------------------
unit_id       dept_id          no_of_emp
----------------------------------------------
1             101              5000
2             102              3000
1             103              4000
1             104              2000
2             105              1000
2             101              3000
----------------------------------------------

Required output: A dynamic view or select:
---------------------------------------------------------------------------
unit_id  unit_loc   ABC-AI  ABC-BI   ABC-CS   XYZ-Testing   XYZ-Development
---------------------------------------------------------------------------
1        Delhi      5000             4000     2000                         
2        Mumbai     3000     3000                           1000
---------------------------------------------------------------------------

问题是ref_department中的每个新部门都对应于view / select查询中的新列。

我写了下面的查询：

variable DEPARTMENTS VARCHAR2(100)
BEGIN
:DEPARTMENTS :=NULL;
  FOR cur_rec IN (SELECT DISTINCT DEPT_DESC FROM REF_DEPARTMENT) LOOP
    :DEPARTMENTS := :DEPARTMENTS || ''''|| cur_rec.DEPT_DESC|| '''' || ',' ;
  END LOOP;
  dbms_output.put_line(rtrim(:DEPARTMENTS,',')); 
select * from
(select uid,uloc,uyear, depdesc, emp from 
(select C.unit_id as uid, C.unit_loc as uloc,C.year as uyear, E.DEPT_ID as depid,R.DEPT_NAME as depname,R.DEPT_DESC as depdesc,S.NO_OF_EMP as emp
FROM Unit U INNER JOIN Employees E ON U.Unit_id=E.unit_id INNER JOIN REF_DEPARTMENT R ON  E.DEPT_ID=R.DEPT_ID))
pivot
(min(emp) for depdesc in (:departments));
END;

错误：PL / SQL：ORA-56901：不允许使用非常量表达式| unpivot

我在下面提到了链接： Pivoting rows into columns dynamically in Oracle; http://www.orafaq.com/forum/t/187172/

Answer 1

很久以前（在Oracle的PIVOT查询存在之前），我编写了一个blog post on "Pivot" Queries，它为您提供了一个包的代码，以帮助构建此类查询。

根据您的要求，您可以这样做：

declare
   rc sys_refcursor;
begin
   rc := pivot.pivot_cursor
               ( group_cols   => 'u.unit_id, u.unit_loc'
               , pivot_col    => 'rd.dept_desc'
               , tables       => 'CompanyUnit x, Employees e, ref_departmnent rd'
               , value_cols   => 'e.no_of_emp'
               , agg_types    => 'sum'
               , where_clause => 'e.dept_id = rd.dept_id and e.unit_id = c.unit_id' 
               );
end;
/

由于ref游标可以具有可变数量的列，因此您需要使用DBMS_SQL包来解析它，找出列名并运行它。起点是将ref游标转换为DBMS_SQL游标：

n := dbms_sql.to_cursor_number(rc);

您需要参考DBMS_SQL程序包文档示例才能更进一步。

在oracle数据库中动态地旋转行

1 个答案: