在 postgres db 中加入和过滤 json 文本字段

Question

我正在使用 postgres v.10 数据库

因此，在我的示例中，我需要归还所有空置但有电的建筑物。

我有一张桌子，里面有所有的建筑，我可以过滤掉空的建筑。

我的问题是有关 elektricity 的信息在成本表中，而我需要的信息存储在 json 格式的文本字段中。它可能看起来像这样：

{"splitkey_hash":{"100":[{"part":"0.0","notes":"blabla","uid":"100_0","id":"100","active_from":"2020-01-01"},{"part":"1.0","notes":"roar","uid":"100_1","id":"100","active_from":"2020-06-01"},{"part":"1.0","notes":null,"uid":"100_2","id":"100","active_from":"2021-01-01"}],"200":[{"part":"0.0","notes":null,"uid":"200_0","id":"200","active_from":"2015-01-01"}],"300":[{"part":"1.0","notes":null,"uid":"300_0","id":"300","active_from":"2013-01-01"}],"400":[{"part":"1.0","notes":"abc","uid":"400_0","id":"400","active_from":"2019-01-01"},{"part":"1.0","notes":null,"uid":"400_1","id":"400","active_from":"2020-09-01"}]}}

JSON 格式：

{
  "splitkey_hash": {
    "100": [
      {
        "part": "0.0",
        "notes": "blabla",
        "uid": "100_0",
        "id": "100",
        "active_from": "2020-01-01"
      },
      {
        "part": "1.0",
        "notes": "roar",
        "uid": "100_1",
        "id": "100",
        "active_from": "2020-06-01"
      },
      {
        "part": "1.0",
        "notes": null,
        "uid": "100_2",
        "id": "100",
        "active_from": "2021-01-01"
      }
    ],
    "200": [
      {
        "part": "0.0",
        "notes": null,
        "uid": "200_0",
        "id": "200",
        "active_from": "2015-01-01"
      }
    ],
    "300": [
      {
        "part": "1.0",
        "notes": null,
        "uid": "300_0",
        "id": "300",
        "active_from": "2013-01-01"
      }
    ],
    "400": [
      {
        "part": "1.0",
        "notes": "abc",
        "uid": "400_0",
        "id": "400",
        "active_from": "2019-01-01"
      },
      {
        "part": "1.0",
        "notes": null,
        "uid": "400_1",
        "id": "400",
        "active_from": "2020-09-01"
      }
    ]
  }
}

---

Building_id = id (json)
Elektricity active = "part": "1.0" (json)
Not active = "part": "0.0" (json)

因此，如果我想问一下 buildung 100 是否在 2020 年 2 月 1 日启用了 elektricity，答案应该是否定的。但是在 2020-06-01 之后会是。

希望你们不明白我在寻找什么？我想以某种方式将 json 中的信息与我所有的 ID 加入到空 buildungs 中，并使用日期进行过滤。

Answer 1

假设建筑物 id 的类型为整数，那么以下查询将返回给定 ref_date 的 active = true 或 false 的所有空建筑物：

来自 PostgreSQL v12：

SELECT DISTINCT ON ( o->'id')
       o->'id'
     , first_value(o->>'part') OVER(PARTITION BY o->'id' ORDER BY (o->>'active_from') :: date DESC) = '1.0' AS active -- = true if active, false if not active
  FROM (your_empty_building_id_list) AS l
 INNER JOIN 
     ( your_cost_table AS t
       CROSS JOIN LATERAL jsonb_path_query((t.your_json_column :: jsonb)->'splitkey_hash'
, '$.*[*]') AS o
     )
    ON l.id = (o->>'id') :: integer
WHERE (o->>'active_from') :: date <= ref_date -- ref_date to be replaced by the effective date used for the query

在 PostgreSQL v12 之前：

SELECT DISTINCT ON ( o->'id')
       o->'id'
     , first_value(o->>'part') OVER(PARTITION BY o->'id' ORDER BY (o->>'active_from') :: date DESC) = '1.0' AS active -- = true if active, false if not active
  FROM (your_empty_building_id_list) AS l
 INNER JOIN 
     ( your_cost_table AS t
       CROSS JOIN LATERAL jsonb_each((t.your_json_column :: jsonb)->'splitkey_hash') AS m(key, value)
       CROSS JOIN LATERAL jsonb_array_elements(m.value) AS o
     ) ON l.id = o->>'id'
    WHERE (o->>'active_from') :: date <= ref_date -- ref_date to be replaced by the effective date used for the query

在dbfiddle查看测试结果