SQL中树结构的元素总和

Question

我有两个表，定义如下：

CREATE TABLE Parts (
   id INTEGER PRIMARY KEY, 
   Name TEXT, 
   Quantity INTEGER, 
   Parentid INTEGER
);  

CREATE TABLE Weight(
   Name TEXT, 
   Weight INTEGER
);

部件表包含父部件和子部件的行。 对于产品 A，有几个部分（B1 和 B2）和子部分（C1-C4）：
内容类似于：
答：2xB1、2xB2、1xC1、5xC2
B1：3xC1、2xC2
B2：1xB1，1xC3

重量如下：
C1：5
C2：4
C3：2

...或：

╔════════════════════════════════════╗
║ Parts                              ║
╠═══════╦══════╦══════════╦══════════╣
║ id    ║ Name ║ Quantity ║ Parentid ║
╠═══════╬══════╬══════════╬══════════╣
║ A     ║ A    ║ 1        ║ NULL     ║
║ B1    ║ B1   ║ 2        ║ A        ║
║ B2    ║ B2   ║ 2        ║ A        ║
║ C1    ║ C1   ║ 1        ║ A        ║
║ C2    ║ C2   ║ 5        ║ A        ║
║ C1    ║ C1   ║ 3        ║ B1       ║
║ C2    ║ C2   ║ 2        ║ B1       ║
║ B1    ║ B1   ║ 1        ║ B2       ║
║ C3    ║ C3   ║ 1        ║ B2       ║
╚═══════╩══════╩══════════╩══════════╝

╔═════════════════╗
║ Weight          ║
╠════════╦════════╣
║ Name   ║ Weight ║
╠════════╬════════╣
║ C1     ║ 5      ║
║ C2     ║ 4      ║
║ C3     ║ 2      ║
║ C4     ║ 8      ║
╚════════╩════════╝

表格中是子零件 C1-C4 的重量和每个零件的数量。 如何获得每个零件和整个产品的重量？

我知道这是递归 CTE，但我无法得到我想要的结果。 我在Recursive sum in tree structure 的帮助下进行了尝试，但没有取得多大成功。

这就是我所拥有的：

WITH c as (
SELECT Parts.id, 
  Parts.quantity, 
  Parts.id as RootID, 
  (Parts.quantity * Weight.weight) as Weight 
FROM Parts LEFT JOIN Weight ON Weight.name = Parts.name
UNION ALL
SELECT Parts.id, 
  Parts.quantity, 
  c.RootID, 
  (Parts.quantity * Weight.weight) as Weight
FROM Parts LEFT JOIN Weight on Weight.name = Parts.name
  INNER JOIN c on Parts.Parentid = c.id
)
SELECT Parts.id, 
  Parts.parent, 
  Parts.name, 
  Parts.quantity, 
  S.SumWeight
FROM Parts INNER JOIN (
  SELECT rootid, 
    SUM(weight) as SumWeight
  FROM c 
  GROUP BY rootid
) as S on Parts.id = S.rootid 
ORDER BY Parts.id

我想要达到的结果是这样的：
121
B1 23
B2 25

感谢任何帮助！

Answer 1

尚不清楚在您的设计中应如何对权重求和以避免重复包含权重，但这是使用与您建议的表格非常相似的表格进行递归的示例。

CREATE TABLE Parts (
   Id varchar(20), 
   Name varchar(20), 
   Quantity INTEGER, 
   ParentId varchar(20)
);  

CREATE TABLE Weight(
   Name varchar(20), 
   [Weight] INTEGER
);

insert into Parts (Id, Name, Quantity, Parentid)
values ('A','A', 1, NULL),
('B1','B1', 2, 'A'),
('B2','B2', 2, 'A'),
('C1','C1', 1, 'A'),
('C2','C2', 5, 'A'),
('C1','C1', 3, 'B1'),
('C2','C2', 2, 'B1'),
('B1','B1', 1, 'B2'),
('C3','C3', 1, 'B2');

insert into Weight(Name, Weight)
values ('C1', 5),
('C2', 4),
('C3', 2),
('C4', 8);


WITH a as (
  SELECT x.Id as EndId,
    x.Id as ComponentId,
    0 as ComponentQuantity,
    1 as Lvl
  FROM Parts x INNER JOIN Parts y ON x.Id = y.ParentId -- must have at least 1 child...
  GROUP BY x.Id
  UNION ALL
  SELECT a.EndId,
    x.Id as ComponentId,
    x.Quantity as ComponentQuantity,
    a.Lvl + 1 as Lvl
  FROM Parts x INNER JOIN a ON x.ParentId = a.ComponentId
)
SELECT a.EndId,
  SUM(a.ComponentQuantity) as SumQuantity,
  SUM(a.ComponentQuantity * b.Weight) as SumWeight
FROM a INNER JOIN Weight b ON a.ComponentId = b.Name
WHERE a.EndId <> a.ComponentId
    AND a.Lvl = 2 -- treat 1st child as the weighted subassembly of interest
GROUP BY a.EndId;