PostgreSQL JSONB - 具有变量键名的查询条件

Question

我已经阅读了各种JSONB教程:

• https://blog.codeship.com/unleash-the-power-of-storing-json-in-postgres/
• https://www.wagonhq.com/sql-tutorial/values-from-nested-json
• http://schinckel.net/2014/05/25/querying-json-in-postgres/
• http://stormatics.com/howto-use-json-functionality-in-postgresql/

请考虑以下示例.

有一张叫做的桌子plans.它包含以下列:

1. id (整数,自动递增主键).
2. name (串).
3. structure (jsonb).

结构列具有常规JSON对象,具有以下结构:

{
  "some_unique_id": {
    "key1": "valueA",   // Fixed key name.
    "key2": "valueB"    // Fixed key name.
  },
  "another_unique_id": {
    "key1": "valueC",   // Fixed key name.
    "key2": "valueB"    // Fixed key name.
  },
  ...                   // can go on up to a 1000 items.
}

注意:最外面的键是动态的.他们改变了每个项目. 这些值只是常规的JSON对象.没什么特别的.

我使用UUID作为结构中的键,因此如果我知道它的UUID,它很容易查找和检索特定值.

另一个选项是使我的结构成为一个对象数组(并将UUID作为值放在每个对象中),如下所示:

[
  {
    "uuid": "some_unique_id",
    "key1": "valueA",   // Fixed key name.
    "key2": "valueB"    // Fixed key name.
  },
  {
    "uuid": "another_unique_id",
    "key1": "valueC",   // Fixed key name.
    "key2": "valueB"    // Fixed key name.
  },
  ...                   // can go on up to a 1000 items.
]

在后一种方法中,要使用其UUID检索特定对象,我必须遍历整个数组并匹配uuid每个对象的键.

所以,我选择了第一种方法.

该表有3条记录.对于这个问题,id和name列的值并不重要.

structure3条记录中列的实际值如下.

记录1:

{
  "bab6246d-802c-4b80-af41-ab15fd1541b4": {
    "name": "Sanskrit",
    "children_uuids": [
      "fa42b4b2-a958-42f1-af33-314e8e1fb6a6",
      "3aeeadfe-6ad4-4229-85a5-5de030c08014"
    ],
    "is_invisible_node": true,
    "tags": [
      "paper",
      "course_paper"
    ],
    "type": "course_paper"
  },
  "dbc33473-8453-4cf9-8ecf-d8013283b0d8": {
    "name": "French",
    "children_uuids": [
      "4bf65ff9-3b11-42d5-a744-adcd1fd5a953"
    ],
    "is_invisible_node": true,
    "tags": [
      "paper",
      "course_paper"
    ],
    "type": "course_paper"
  }
}

记录2:

{
  "ed6164d0-fdc0-4259-90a5-fd60d9d716dc": {
    "name": "Pen and Paper Assessment 1",
    "children_uuids": [

    ],
    "is_invisible_node": false,
    "tags": [
      "paper",
      "assessment"
    ],
    "type": "assessment"
  },
  "059d0116-bca2-49f1-b333-58c4dbec8566": {
    "name": "Content",
    "children_uuids": [

    ],
    "is_invisible_node": false,
    "tags": [
      "paper",
      "assessment"
    ],
    "type": "assessment"
  }
}

记录3:

{
  "63619c7f-fa73-49af-9df5-4be1eb38cee5": {
    "name": "Q12",
    "children_uuids": [

    ],
    "is_invisible_node": true,
    "tags": [
      "paper",
      "regular_paper"
    ],
    "type": "regular_paper"
  },
  "56eed164-17f7-48e9-b3ce-b5b469e8cb0e": {
    "name": "Q13",
     "children_uuids": [

    ],
    "is_invisible_node": false,
    "tags": [
      "paper",
      "regular_paper"
    ],
    "type": "regular_paper"
  },
  "69d202c1-5c23-412f-860d-1a5d705c31b3": {
    "name": "Q14",
    "children_uuids": [

    ],
    "is_invisible_node": false,
    "tags": [
      "paper",
      "regular_paper"
    ],
    "type": "regular_paper"
  }
}

现在,如何编写查询来执行以下两项操作？

• 我想获取包含属性设置为true的任何对象的所有记录is_invisible_node.
• 我想获得包含其中一个的所有对象.regular_papertags

感谢您阅读这篇文章!任何帮助,将不胜感激.

Answer 1

我想获取包含is_invisible_node属性设置为true的任何对象的所有记录.

使用jsonb_each()检索第二级别对象:

select id, uuid.key uuid
from 
    plans, 
    jsonb_each(structure) uuid
where (value->>'is_invisible_node')::boolean;

 id |                 uuid                 
----+--------------------------------------
  1 | bab6246d-802c-4b80-af41-ab15fd1541b4
  1 | dbc33473-8453-4cf9-8ecf-d8013283b0d8
  3 | 63619c7f-fa73-49af-9df5-4be1eb38cee5
(3 rows)

要么

select distinct id
from 
    plans, 
    jsonb_each(structure) uuid
where (value->>'is_invisible_node')::boolean;

 id 
----
  1
  3
(2 rows)    

我希望将包含regular_paper的所有对象作为其标记之一.

json对象tags是一个数组,因此不需要jsonb_array_elements_text():

select uuid.key uuid
from 
    plans, 
    jsonb_each(structure) uuid,
    jsonb_array_elements_text(value->'tags') tag
where tag = 'regular_paper';

                 uuid                 
--------------------------------------
 56eed164-17f7-48e9-b3ce-b5b469e8cb0e
 63619c7f-fa73-49af-9df5-4be1eb38cee5
 69d202c1-5c23-412f-860d-1a5d705c31b3
(3 rows)