如何使用 sql 获取树中所有父母的孩子的累积总和？

Question

so I have a sql query to get the tree structure with unlimited depth as follows. But i'm clueless about getting the total for each parent.

I have a table dimensionValue which has id, dimensionValueId, name I have another table valuation which has id, dimensionValue and volume. The volume needs to be added at each parent level.

I have http://sqlfiddle.com/#!15/5c0e7/5

I also want to get the flat sql with cumulative total, Currently I only have tree.

WITH RECURSIVE hierarchy AS (
    SELECT
        id,
        name,
        dimensionvalueid,
        0 AS level
    FROM
        dimensionvalue d
    WHERE
        dimensionvalueid IS NULL
    
UNION ALL
SELECT
    e.id,
    e.name,
    e.dimensionvalueid,
    hierarchy.level + 1
FROM
    dimensionvalue e,
    hierarchy
WHERE
    e.dimensionvalueid = hierarchy.id
)
SELECT
    d.id, d.dimensionvalueid, d.name, v.volume
FROM
    hierarchy d
    LEFT JOIN valuation v ON v.dimensionvalueid = d.id
        
    GROUP BY
        d.id,
        d.dimensionvalueid,
        d.name,
        v.volume
    ORDER BY
        d.id;

I have gotten the tree but I want to get the cumulative sum of volume from bottom up to all parent nodes. In my application only lower level volume are added which gets add up to the parent.

Later I want to nest the flat data as follows: The output I want is:

   [
  {
    "id": 1,
    "name": "A1",
    "dimensionValueId": null,
    "sumVolume": 300,
    "children": [
      {
        "id": 2,
        "name": "A1:1",
        "dimensionValueId": 1,
        "sumVolume": 300,
        "children": [
          {
            "id": 3,
            "name": "A1:1:1",
            "dimensionValueId": 2,
            "sumVolume": 300,
            "children": [
              {
                "id": 4,
                "name": "A1:1:1:1",
                "dimensionValueId": 3,
                "sumVolume": 200,
                "children": null
              },
              {
                "id": 5,
                "name": "A1:1:1:2",
                "dimensionValueId": 3,
                "sumVolume": 100,
                "children": null
              }
            ]
          }
        ]
      }
    ]
  },
  {
    "id": 6,
    "name": "B1",
    "dimensionValueId": null,
    "sumVolume": 200,
    "children": [
      {
        "id": 7,
        "name": "B1:1",
        "dimensionValueId": 6,
        "sumVolume": 100,
        "children": null
      },
      {
        "id": 8,
        "name": "B1:2",
        "dimensionValueId": 6,
        "sumVolume": 100,
        "children": [
          {
            "id": 9,
            "name": "B1:2:1",
            "dimensionValueId": 6,
            "sumVolume": 100,
            "children": null
          }
        ]
      }
    ]
  }
]

Answer 1

Using a recursive cte :

with recursive cte(id, h) as (
   select d.id, ('['||d.id||']')::jsonb from dimensionvalue d where d.dimensionvalueid is null
   union all
   select d.id, c.h || ('['||c.id||']')::jsonb from cte c 
   join dimensionvalue d on d.dimensionvalueid = c.id
),
cum_sum(id, s) as (
   select d.id, sum(v1.volume) from dimensionvalue d 
   join cte c on (exists (select 1 from jsonb_array_elements(c.h) v where v.value::int = d.id) 
              and not exists (select 1 from dimensionvalue d1 where d1.dimensionvalueid = c.id)) 
   join valuation v1 on v1.dimensionvalueid = c.id group by d.id
),
tree(id, p, js) as (
   select d.id, d.dimensionvalueid, 'null'::json from dimensionvalue d where not exists (
       select 1 from dimensionvalue d1 where d1.dimensionvalueid = d.id)
   union all
   select d4.id, d4.p, json_agg(d4.js) js from (
       select d.id id, d.dimensionvalueid p, json_build_object('id', t.id, 'name', d3.name, 
            'dimensionValueId', d3.dimensionvalueid, 
            'sumVolume', coalesce(c.s, v1.volume), 'children', t.js) js 
       from dimensionvalue d join tree t on d.id = t.p 
       left join cum_sum c on c.id = t.id join valuation v1 on v1.dimensionvalueid = t.id 
       join dimensionvalue d3 on d3.id = t.id) d4 group by d4.id, d4.p
)
select jsonb_pretty(t3.js::jsonb) from (
   select json_agg(json_build_object('id', t.id, 'name', d3.name, 'dimensionValueId', d3.dimensionvalueid, 'sumVolume', c.s, 'children', t.js)) js 
   from (select t1.id, json_agg(v.value) js 
         from tree t1 cross join json_array_elements(t1.js) v 
         where t1.p is null group by t1.id) t 
   join cum_sum c on c.id = t.id join dimensionvalue d3 on d3.id = t.id) t3

See fiddle .