分层查询 - 限制每个级别的元素数量

Question

我正在使用 Oracle DBMS，我有一个关于分层查询的问题。

我正在使用分层 SQL 查询为我的部门创建一个组织列表。组织结构分为三个级别，依次是部门经理、部门经理和团队成员。

到目前为止我的查询是：

SELECT level, employee_number, name, manager, department, phone
FROM employee_table
START WITH manager is null
CONNECT BY PRIOR employee_number = manager;

此查询生成一个层次结构列表，其中部门经理位于第一级，每个部门经理位于第二级，团队成员位于第三级。总的来说，只有一名部门经理，三名科室经理，30名团队成员。

问题：我的要求是将列表中每个级别的员工人数限制为最多三名员工。前两个级别没有问题，因为只有一个部门经理和三个科长（永远不会超过三个科长），但我目前在三级有30个团队成员（每个科长管理10个团队成员） ）。我的目标是3级3个队员，4级3个队员，5级3个队员，以此类推。队员的顺序不重要，所以不管哪个队员在3 级，哪些团队成员处于 4 级等。

为了实现这个目标，我宁愿避免将团队成员的经理设置为另一个团队成员的员工编号。我可以在employee_table 中创建另一个名为“org_list_parent”的列，并指出一个团队成员的“org_list_parent”是另一个团队成员的employee_number，但如果可能的话，我更愿意避免这样做。

有人对这个问题有任何想法吗？

提前谢谢你。

更新：

使用 mathguy 的查询，我能够获得最接近我想要实现的输出。但是，如果可能的话，我想对输出进行一些调整。我正在使用此列表在 Oracle Apex 中构建组织结构图，并使用 mathguy 的表和查询，得到以下输出：

这与我试图制作的视觉效果非常接近。 “每级 3 名团队成员”背后的原因是为了防止图表在水平方向上过大。但是，如果您向左看，例如，员工 1104、1105 和 1106 在 1103 之下，员工 1107 在 1106 之下。最好是员工 1104 在 1101 之下，员工 1105 在 1102 之下，员工 1106 在 1103 之下，和员工 1107 在 1104 下。有什么方法可以编辑查询，以便直观地产生结果？

更新 2：

alexgibbs 要求就他提出的两个不同查询提供反馈，作为我的问题的解决方案。以下是他的第一个查询：

SELECT CASE WHEN LEVEL < 3
          THEN LEVEL
        ELSE 3 + FLOOR((DENSE_RANK() OVER (PARTITION BY MANAGER ORDER BY 
EMPLOYEE_NUMBER ASC ) - 1) / 3) END AS ADJUSTED_LEVEL,
EMPLOYEE_NUMBER, MANAGER
FROM EMPLOYEE_TABLE
START WITH MANAGER IS NULL
CONNECT BY PRIOR EMPLOYEE_NUMBER = MANAGER
ORDER BY 2 ASC, 1 ASC;

以下是此查询在 Oracle Apex 中的列表输出：

以下是他的第二个查询：

SELECT
CASE WHEN LEVEL < 3
 THEN LEVEL
 ELSE 3 + FLOOR((DENSE_RANK() OVER (PARTITION BY MANAGER ORDER BY 
EMPLOYEE_NUMBER ASC ) - 1) / 3) END AS ADJUSTED_LEVEL,
EMPLOYEE_NUMBER,
MANAGER,
DEPARTMENT
FROM EMPLOYEE_TABLE
START WITH MANAGER IS NULL
CONNECT BY PRIOR EMPLOYEE_NUMBER = MANAGER
ORDER BY
NVL2(MANAGER,1,0) ASC,
DEPARTMENT ASC,
CASE WHEN LEVEL < 3
 THEN LEVEL
 ELSE (MOD((DENSE_RANK() OVER (PARTITION BY MANAGER ORDER BY EMPLOYEE_NUMBER 
ASC )) - 1,3) + 3) END ASC,
ADJUSTED_LEVEL ASC;

以下是此查询在 Oracle Apex 中的输出：

Answer 1

我会这样做。

首先是测试数据（我包括了足够多的员工来说明这一点，但每个中层管理人员的人数还不到 10 人）。由于与手头的问题无关，我省略了电话号码。

create table employee_table(employee_number, name, manager, department) as
    select 1001, 'Big Boss', null, 100 from dual union all
    select 1100, 'Beth Mgr', 1001, 100 from dual union all
    select 1101, 'Jim'     , 1100, 100 from dual union all
    select 1102, 'Jackie'  , 1100, 100 from dual union all
    select 1103, 'Helen'   , 1100, 100 from dual union all
    select 1104, 'Tom'     , 1100, 100 from dual union all
    select 1105, 'Vance'   , 1100, 100 from dual union all
    select 1106, 'Rosa'    , 1100, 100 from dual union all
    select 1107, 'Chuck'   , 1100, 100 from dual union all
    select 1200, 'Duck Mgr', 1001, 200 from dual union all
    select 1201, 'Danny'   , 1200, 200 from dual union all
    select 1202, 'Henry'   , 1200, 200 from dual union all
    select 1203, 'Mac'     , 1200, 200 from dual union all
    select 1204, 'Hassan'  , 1200, 200 from dual union all
    select 1205, 'Ann'     , 1200, 200 from dual union all
    select 1300, 'Adam Mgr', 1001, 300 from dual union all
    select 1301, 'Wendy'   , 1300, 300 from dual
;

然后是查询。我希望输出遵循“层次顺序”（如果我们不必弄乱级别的话）。为此，我首先运行分层查询，捕获ROWNUM 以对最终结果进行排序，然后修改外部查询中的级别。请注意，我使用LVL 作为列名； LEVEL 是保留字，所以不能用作列名。

select   case lvl when 3 then lvl + ceil(rn/3) - 1 else lvl end as lvl,
         employee_number, name, manager, department
from     (
          select     level as lvl, employee_number, name, manager, department,
                     rownum as ord, 
                     row_number() over 
                        (partition by manager order by employee_number) as rn
          from       employee_table
          start with manager is null
          connect by prior employee_number = manager
         )
order by ord
;

输出：

       LVL EMPLOYEE_NUMBER NAME        MANAGER DEPARTMENT
---------- --------------- -------- ---------- ----------
         1            1001 Big Boss                   100
         2            1100 Beth Mgr       1001        100
         3            1101 Jim            1100        100
         3            1102 Jackie         1100        100
         3            1103 Helen          1100        100
         4            1104 Tom            1100        100
         4            1105 Vance          1100        100
         4            1106 Rosa           1100        100
         5            1107 Chuck          1100        100
         2            1200 Duck Mgr       1001        200
         3            1201 Danny          1200        200
         3            1202 Henry          1200        200
         3            1203 Mac            1200        200
         4            1204 Hassan         1200        200
         4            1205 Ann            1200        200
         2            1300 Adam Mgr       1001        300
         3            1301 Wendy          1300        300

Answer 2

注意：已针对问题中的更新进行了重大编辑，并且不确定/未与 apex 进行兼容性测试

下面的解决方案只会在级别 2 的第三个之外添加部分经理。但是对于级别 3+，如果在部分经理之外如何分配其他级别并不重要，则可以根据 employee_id 分配它们。更新后，三人组的每个成员都应将其伪下一级直接嵌套在下面，下面是一个示例，它根据员工 ID 在三人组中分配伪级别和子级别，部门。

下面是一个示例数据集和查询。

CREATE TABLE EMPLOYEE_TABLE (
  EMPLOYEE_NUMBER NUMBER,
  MANAGER NUMBER DEFAULT NULL,
  NAME CHARACTER VARYING(64 BYTE),
  DEPARTMENT CHARACTER VARYING(64 BYTE),
  PHONE CHARACTER VARYING(64 BYTE)
);


--Dept Manager
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (10,NULL, 1);
--Section Managers
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (200,10, 2);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (300,10, 3);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (400,10, 4);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (500,10, 5);
-- Section Employees
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (2010,200, 2);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (2020,200, 2);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (2030,200, 2);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (2040,200, 2);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (2050,200, 2);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (2060,200, 2);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (2070,200, 2);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (2080,200, 2);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (2090,200, 2);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (2100,200, 2);

INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (3010,300, 3);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (3020,300, 3);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (3030,300, 3);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (3040,300, 3);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (3050,300, 3);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (3060,300, 3);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (3070,300, 3);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (3080,300, 3);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (3090,300, 3);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (3100,300, 3);

INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (4010,400, 4);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (4020,400, 4);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (4030,400, 4);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (4040,400, 4);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (4050,400, 4);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (4060,400, 4);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (4070,400, 4);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (4080,400, 4);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (4090,400, 4);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (4100,400, 4);


INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (5010,500, 5);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (5020,500, 5);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (5030,500, 5);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (5040,500, 5);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (5050,500, 5);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (5060,500, 5);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (5070,500, 5);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (5080,500, 5);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (5090,500, 5);
INSERT INTO EMPLOYEE_TABLE(EMPLOYEE_NUMBER, MANAGER, DEPARTMENT) VALUES (5100,500, 5);
COMMIT;

已编辑 查询：

SELECT
CASE WHEN LEVEL < 3
     THEN LEVEL
     ELSE 3 + FLOOR((DENSE_RANK() OVER (PARTITION BY MANAGER ORDER BY EMPLOYEE_NUMBER ASC ) - 1) / 3) END AS ADJUSTED_LEVEL,
EMPLOYEE_NUMBER,
MANAGER,
DEPARTMENT
FROM EMPLOYEE_TABLE
START WITH MANAGER IS NULL
CONNECT BY PRIOR EMPLOYEE_NUMBER = MANAGER
ORDER BY
NVL2(MANAGER,1,0) ASC,
DEPARTMENT ASC,
CASE WHEN LEVEL < 3
     THEN LEVEL
     ELSE (MOD((DENSE_RANK() OVER (PARTITION BY MANAGER ORDER BY EMPLOYEE_NUMBER ASC )) - 1,3) + 3) END ASC,
ADJUSTED_LEVEL ASC;

结果：

  ADJUSTED_LEVEL   EMPLOYEE_NUMBER   MANAGER DEPARTMENT
               1                10           1
               2               200        10 2
               3              2010       200 2
               4              2040       200 2
               5              2070       200 2
               6              2100       200 2
               3              2020       200 2
               4              2050       200 2
               5              2080       200 2
               3              2030       200 2
               4              2060       200 2
               5              2090       200 2
               2               300        10 3
               3              3010       300 3
               4              3040       300 3
               5              3070       300 3
               6              3100       300 3
               3              3020       300 3
               4              3050       300 3
               5              3080       300 3
               3              3030       300 3
               4              3060       300 3
               5              3090       300 3
               2               400        10 4
               3              4010       400 4
               4              4040       400 4
               5              4070       400 4
               6              4100       400 4
               3              4020       400 4
               4              4050       400 4
               5              4080       400 4
               3              4030       400 4
               4              4060       400 4
               5              4090       400 4
               2               500        10 5
               3              5010       500 5
               4              5040       500 5
               5              5070       500 5
               6              5100       500 5
               3              5020       500 5
               4              5050       500 5
               5              5080       500 5
               3              5030       500 5
               4              5060       500 5
               5              5090       500 5