如何连接或 Listagg 同一列中的两个值以将其与其他列中的值连接

Question

我有两个表:

TABLE_A

pos_id    res_id     bb_id    bsk_name
1122      10000       1444     type_1 
1122      10000       5678     type_2
1122      10001       1444     type_1
1122      10001       5678     type_2
1122      10002       1467     type_1
1122      10002       5678     type_2
1122      10003       1467     type_1
1122      10003       5678     type_2

table_b

pos_id    row_id    bb_id     bsk_name
1122        1        1444      type_1
1122        1        5678      type_2
1122        2        1467      type_1
1122        2        5678      type_2

我想加入 table_a 和 table_b 以获得每个 res_id 的 row_id。

res_id 10000 和 10001 必须有 row_id 1 而 res_id 10002 和 10003 必须有 row_id 2。但是因为没有唯一的连接这两个表的列我得到了 bb_id 5678 的重复值，因为它们在 row_id 中是相同的。

所以有没有办法像在 table_a 中的 erg_id 和 table_b 中的 row_id 一起列出 bb_id 来加入这两个表格 ?

Answer 1

您可以使用 CAST(COLLECT(...) AS ...) 将行聚合到用户定义的集合中，然后比较这些集合。

首先，创建一个嵌套-table-类型的集合：

CREATE TYPE int_list AS TABLE OF INT;

那么你可以使用：

SELECT a.pos_id,
       a.res_id,
       a.bb_id,
       a.bsk_name,
       b.row_id,
       b.bsk_name
FROM   (
  SELECT a.*,
         CAST(
           COLLECT(bb_id) OVER (PARTITION BY pos_id, res_id)
           AS int_list
         ) AS all_bb_ids
  FROM   table_a a
) a
INNER JOIN (
  SELECT b.*,
         CAST(
           COLLECT(bb_id) OVER (PARTITION BY pos_id, row_id)
           AS int_list
         ) AS all_bb_ids
  FROM   table_b b
) b
ON (    a.pos_id = b.pos_id
    AND a.all_bb_ids = b.all_bb_ids
    AND a.bb_id = b.bb_id)

其中，对于您的示例数据：

CREATE TABLE TABLE_A ( pos_id, res_id, bb_id, bsk_name ) AS
SELECT 1122, 10000, 1444, 'type_1' FROM DUAL UNION ALL
SELECT 1122, 10000, 5678, 'type_2' FROM DUAL UNION ALL
SELECT 1122, 10001, 1444, 'type_1' FROM DUAL UNION ALL
SELECT 1122, 10001, 5678, 'type_2' FROM DUAL UNION ALL
SELECT 1122, 10002, 1467, 'type_1' FROM DUAL UNION ALL
SELECT 1122, 10002, 5678, 'type_2' FROM DUAL UNION ALL
SELECT 1122, 10003, 1467, 'type_1' FROM DUAL UNION ALL
SELECT 1122, 10003, 5678, 'type_2' FROM DUAL;

CREATE TABLE table_b (pos_id, row_id, bb_id, bsk_name) AS
SELECT 1122, 1, 1444, 'type_1' FROM DUAL UNION ALL
SELECT 1122, 1, 5678, 'type_2' FROM DUAL UNION ALL
SELECT 1122, 2, 1467, 'type_1' FROM DUAL UNION ALL
SELECT 1122, 2, 5678, 'type_2' FROM DUAL;

输出：

POS_ID	RES_ID	BB_ID	BSK_NAME	ROW_ID	BSK_NAME
1122	10000	1444	type_1	1	type_1
1122	10000	5678	type_2	1	type_2
1122	10001	1444	type_1	1	type_1
1122	10001	5678	type_2	1	type_2
1122	10002	1467	type_1	2	type_1
1122	10002	5678	type_2	2	type_2
1122	10003	1467	type_1	2	type_1
1122	10003	5678	type_2	2	type_2

db<>fiddle here

---

更新

您也可以使用 LISTAGG 来聚合数据；但是，如果聚合字符串超过 4000 字节，则聚合将失败（使用 CAST(COLLECT(...) ...) 没有此限制，但需要您创建集合数据类型）：

SELECT a.pos_id,
       a.res_id,
       a.bb_id,
       a.bsk_name,
       b.row_id,
       b.bsk_name
FROM   (
  SELECT a.*,
         LISTAGG(bb_id, ',') WITHIN GROUP (ORDER BY bb_id)
           OVER (PARTITION BY pos_id, res_id) AS all_bb_ids
  FROM   table_a a
) a
INNER JOIN (
  SELECT b.*,
         LISTAGG(bb_id, ',') WITHIN GROUP (ORDER BY bb_id)
           OVER (PARTITION BY pos_id, row_id) AS all_bb_ids
  FROM   table_b b
) b
ON (    a.pos_id = b.pos_id
    AND a.all_bb_ids = b.all_bb_ids
    AND a.bb_id = b.bb_id)

db<>fiddle here