查询 Rails 中的 Postgres JSON 数组字段

Query on Postgres JSON array field in Rails

我正在尝试查询 Postgres 数据库中的某个值。我在 users table 中有一个名为 groups 的字段,可以用以下任一方式表示:

1.

groups: {"data"=>[{"serie"=>5, "year"=>3, "specialization"=>"Matematica", "management_id"=>1, "group_number"=>2}, {"serie"=>5, "year"=>3, "specialization"=>"Matematica", "management_id"=>1, "group_number"=>2}]}

2.

groups: [{"serie"=>5, "year"=>3, "specialization"=>"Matematica", "management_id"=>1, "group_number"=>2}, {"serie"=>5, "year"=>3, "specialization"=>"Matematica", "management_id"=>1, "group_number"=>2}]

我对这两种表述都满意。但是,我似乎无法找到如何获得 serie 5 中的所有用户。我尝试了多个查询:

@users = User.where("groups ->> 'data' @>  ?", {serie: 5})
@users = User.where("groups -> 'data' @>  '?'", {serie: 5})
@users = User.where("groups ->> 'data' ->> 'serie' = ?", 5)

还有很多其他的尝试,有些比其他的更愚蠢(见上文)。我该怎么做?

我已经能够确定:

select groups -> 'data' ->> 'serie' from users;  
ERROR: cannot extract field from a non-object.

但是以下查询有效:

select json_array_elements(groups -> 'data') ->> 'serie' from users;

我认为我没有正确提供列中的数据。我提供的哈希是:

pry(#<Overrides::RegistrationsController>)> @response['data']['user']
=> {"last_name"=>"Doe1",
 "first_name"=>"John1",
 "email"=>"c0f45@example.com",
 "groups"=>
  {"data"=>
    [{"serie"=>5, "year"=>3, "specialization"=>"Matematica", "management_id"=>1, "group_number"=>2}, {"serie"=>5, "year"=>3, "specialization"=>"Matematica", "management_id"=>1, "group_number"=>2}]}}

保存资源之前是这样的:

pry(#<Overrides::RegistrationsController>)> @resource
=> #<User id: nil, provider: "email", uid: "", first_name: "John1", last_name: "Doe1", email: "c0f45@example.com", role: "Student", created_at: nil, updated_at: nil, groups: {"data"=>[{"serie"=>5, "year"=>3, "specialization"=>"Matematica", "management_id"=>1, "group_number"=>2}, {"serie"=>5, "year"=>3, "specialization"=>"Matematica", "management_id"=>1, "group_number"=>2}]}>

假设:

  • Postgres 9.4 或更高版本
  • "get all the users that are in serie 5" 应该是指:
    "with at least one array element that contains {"serie": 5}. There may be others."
  • 使用您的第一个较短的数据格式。没有多余的 'data' 键。

简短回答: 使用 jsonb 而不是 json,这很有效:

User.where("groups @> ?", '[{"serie": 5}]')

注意方括号使右边的操作数成为 JSON array.

为什么?

此处突出的误解:data type json is not the same as jsonb.

你没有声明实际的 table 定义,但你后来评论 json 并且问题中有提示:

select json_array_elements(groups -> 'data') ->> 'serie' from users;

json_array_elements() 仅适用于 json,对于 jsonb 必须是 jsonb_array_elements()。但是您尝试使用未为 json:

定义的 jsonb operator @>
groups -> 'data' @>  '?'

operator ->returns与左手输入相同的类型。但是 @> 只为 jsonb 定义,而不是 json.

然后您尝试使用 text 的运算符 @> 作为左侧操作数。 也不可能:

groups ->> 'data' @>  ?

对于各种类型(包括 Postgres 数组),运算符 @> 有变体,但对于 textjson 都没有。

所以,简短的回答:使用 jsonb 而不是 json。这也允许使用非常 高效的索引

  • Index for finding an element in a JSON array

json

对于数据类型 json 你可以使用:

SELECT *
FROM   users u
WHERE  EXISTS (
   SELECT FROM json_array_elements(u.groups) elem 
   WHERE  elem ->> 'serie' = '5'
   );

演示

jsonb:

SELECT *
FROM  (
   VALUES (1, jsonb '[{"serie":5, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}
                    , {"serie":5, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}]')
        , (2,       '[{"serie":7, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}
                    , {"serie":8, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}]')
        , (3,       '[{"serie":9, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}
                    , {"serie":5, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}]')
   ) users(id, groups)
WHERE  groups @> '[{"serie": 5}]';

json:

SELECT *
FROM  (
   VALUES (1, json  '[{"serie":5, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}
                    , {"serie":5, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}]')
        , (2,       '[{"serie":7, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}
                    , {"serie":8, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}]')
        , (3,       '[{"serie":9, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}
                    , {"serie":5, "year":3, "specialization":"Matematica", "management_id":1, "group_number":2}]')
   ) users(id, groups)
WHERE  EXISTS (
   SELECT FROM json_array_elements(users.groups) elem 
   WHERE  elem ->> 'serie'  = '5'
   );