要求:计算叶子结点和其上级的距离
表结构:
"Goods_Category"
(
"Id" bigint,
"ParentId" bigint
);
1.整理叶子结点及其上级的关系,按顺序存储成数组:[叶子结点1, 叶子结点1-上级Id1, 上级Id1的上级Id2...]
WITH T1 AS(WITH RECURSIVE T AS(
--没有下级的就是叶子节点
SELECT ARRAY[GC."ParentId"] || GC."Id" AS "ParentId"
FROM "Goods_Category" AS GC
JOIN "Goods_Category" AS GC1 ON
GC."ParentId" = GC1."Id"
WHERE NOT EXISTS (
SELECT 1
FROM "Goods_Category" AS GC1
WHERE GC."Id" = GC1."ParentId"
)
UNION ALL (
SELECT GC."ParentId" || T."ParentId" AS "ParentId"
FROM T
JOIN "Goods_Category" AS GC ON
T."ParentId"[1] = GC."Id"
JOIN "Goods_Category" AS GC1 ON
GC."ParentId" = GC1."Id"
)
)
SELECT T1."Id",
T."ParentId"
FROM T
JOIN(
WITH T1 AS(
SELECT T."ParentId"[ARRAY_LENGTH(T."ParentId", 1)] AS "Id",
ARRAY_LENGTH(T."ParentId", 1) AS LENGTH,
T."ParentId"
FROM T
)
SELECT T1."Id",
MAX(T1.LENGTH) AS LENGTH
FROM T1
GROUP BY T1."Id"
) AS T1 ON
T."ParentId"[ARRAY_LENGTH(T."ParentId", 1)] = T1."Id"
AND ARRAY_LENGTH(T."ParentId", 1) = T1.LENGTH
ORDER BY T1."Id"
)
SELECT *
FROM T1;
2.循环T1记录
FOR R IN (SELECT * FROM T1) LOOP
ARRAY_LENGTH = (ARRAY_LENGTH(R."ParentId", 1) - 1);
FOR B_INDEX IN 1..ARRAY_LENGTH LOOP
FOR E_INDEX IN (B_INDEX+1)..(ARRAY_LENGTH+1) LOOP
IF NOT EXISTS(SELECT 1 FROM bi_dim_category_closure WHERE parent_id = R."ParentId"[B_INDEX] AND child_id = R."ParentId"[E_INDEX]) THEN
INSERT INTO bi_dim_category_closure
(
parent_id,
child_id,
distance
)
VALUES(
R."ParentId"[B_INDEX],
R."ParentId"[E_INDEX],
(E_INDEX-B_INDEX)
);
END IF;
END LOOP;
END LOOP;
END LOOP;