Postgresql JSON对象和数组查询

一. Postgresql 9.5以下版本

1.1 简单查询(缺陷：数组必须指定下标，不推荐)

1.1.1 模糊查询

SELECT  * FROM "public"."tf_low_data_testUser" WHERE  "address" #>> '{0,name}' like '%bb%'
1

address字段是JSONArray类型，所以在路径中，使用数字索引来访问数组元素，从 0 开始计数。

1.1.2 等值匹配

SELECT  * FROM "public"."tf_low_data_testUser" WHERE "address" #>> '{0,name}' = 'bbb'
1

如果字段是int类型，后面需要添加::int

1.1.3 时间搜索

SELECT  * FROM "public"."tf_low_data_testUser" WHERE  "address" #>> '{0,date}' BETWEEN '2023-08-13' AND '2023-08-17'
1

1.1.4 在列表

SELECT  * FROM "public"."tf_low_data_testUser" WHERE  "address" #>> '{0,name}' IN ('bbb','ccc')
1

1.1.5 包含

SELECT  * FROM "public"."tf_low_data_testUser" WHERE "address" #> '{0,roles,0,roleUsers}' @> '["eee"]'
1

• #>：获取在指定路径的 JSON 对象，路径不存在则返回空。返回类型是json(b)
• #>>：获取在指定路径的 JSON 对象，路径不存在则返回空。返回类型是text

1.2 多层级JSONArray（推荐）

如果表中有一个字段posts，数据结构为

[{
	"name": "aaa",
	"ports": [{
		"port": 443,
		"nickname": "ggg",
		"date": "2023-08-29",
		"address": ["111", "222"]
	}, {
		"port": 80,
		"nickname": "fff",
		"date": "2022-08-29",
		"address": ["333", "444"]
	}]
}, {
	"name": "bbb",
	"ports": [{
		"port": 2443,
		"nickname": "hhh",
		"date": "2021-08-29",
		"address": ["999"]
	}, {
		"port": 280,
		"nickname": "jjj",
		"date": "2020-08-29",
		"address": ["111111"]
	}]
}]
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

1.2.1 模糊查询

查询nickname like '%jj%'

可以看出有两层JSONArray结构

SELECT * FROM "public"."tf_low_data_testUser" WHERE EXISTS (
  SELECT 1
  FROM jsonb_array_elements("ports") as arr1(obj1) 
	CROSS JOIN jsonb_array_elements(obj1->'ports') as arr2(obj2)
  WHERE (obj2->>'nickname') like '%gg%'
);
1
2
3
4
5
6

当该层级类型是数组就添加CROSS JOIN jsonb_array_elements(obj1->'ports') as arr2(obj2)

1.2.2 模糊查询 NOT

SELECT * FROM "public"."tf_low_data_testUser" WHERE NOT EXISTS (
  SELECT 1
  FROM jsonb_array_elements("ports") as arr1(obj1) 
	CROSS JOIN jsonb_array_elements(obj1->'ports') as arr2(obj2)
  WHERE (obj2->>'nickname') like '%gg%'
);
1
2
3
4
5
6

查的是另外三条数据源

1.2.3 等值匹配

SELECT * FROM "public"."tf_low_data_testUser" WHERE EXISTS (
  SELECT 1
  FROM jsonb_array_elements("ports") as arr1(obj1) 
	CROSS JOIN jsonb_array_elements(obj1->'ports') as arr2(obj2)
  WHERE (obj2->>'port')::int = 80
);
1
2
3
4
5
6

如果是数字类型后面需要转换 ::int，因为 ->> 操作符的返回类型是 text

1.2.4 等值匹配 NOT

SELECT * FROM "public"."tf_low_data_testUser" WHERE NOT EXISTS (
  SELECT 1
  FROM jsonb_array_elements("ports") as arr1(obj1) 
	CROSS JOIN jsonb_array_elements(obj1->'ports') as arr2(obj2)
  WHERE (obj2->>'port')::int = 80
);
1
2
3
4
5
6

查的是另外三条数据源

1.2.5 时间搜索

SELECT * FROM "public"."tf_low_data_testUser" WHERE EXISTS (
  SELECT 1
  FROM jsonb_array_elements("ports") as arr1(obj1) 
	CROSS JOIN jsonb_array_elements(obj1->'ports') as arr2(obj2)
  WHERE (obj2->>'date') BETWEEN '2022-08-13' AND '2023-08-17'
);
1
2
3
4
5
6

1.2.6 时间搜索 NOT

查的是另外三条数据源

1.2.7 在列表

SELECT * FROM "public"."tf_low_data_testUser" WHERE EXISTS (
  SELECT 1
  FROM jsonb_array_elements("ports") as arr1(obj1) 
	CROSS JOIN jsonb_array_elements(obj1->'ports') as arr2(obj2)
  WHERE (obj2->>'nickname') IN ('ggg','fff')
);
1
2
3
4
5
6

1.2.8 在列表 NOT

查的是另外三条数据源

1.2.9 包含

SELECT * FROM "public"."tf_low_data_testUser" WHERE EXISTS (
  SELECT 1
  FROM jsonb_array_elements("ports") as arr1(obj1) 
	CROSS JOIN jsonb_array_elements(obj1->'ports') as arr2(obj2)
  WHERE (obj2->'address') @> '["444"]'
);
1
2
3
4
5
6

此时使用的操作符是->，返回值是jsonb类型

1.2.10 包含 NOT

查的是另外三条数据源

二. Postgresql 9.5和以上版本

也兼容上面的JSON查询

2.1 模糊查询

使用函数jsonb_path_exists（可以指定JSON路径，如果是数组添加[*]）的正则查询达到模糊查询的效果

-- like '%ggg%'
SELECT * FROM "public"."tf_low_data_testUser" WHERE jsonb_path_exists("ports", '$[*].ports[*].nickname ?(@ like_regex "g")')
-- 左模糊 like '%g'
SELECT * FROM "public"."tf_low_data_testUser" WHERE jsonb_path_exists("ports", '$[*].ports[*].nickname ?(@ like_regex "g$")')
-- 右模糊 like 'g%'
SELECT * FROM "public"."tf_low_data_testUser" WHERE jsonb_path_exists("ports", '$[*].ports[*].nickname ?(@ like_regex "^g")')
-- 等值匹配
SELECT * FROM "public"."tf_low_data_testUser" WHERE jsonb_path_exists("ports", '$[*].ports[*].nickname ?(@ like_regex "^ggg$")')
1
2
3
4
5
6
7
8

同样支持NOT

2.2 等值匹配

SELECT * FROM "public"."tf_low_data_testUser" WHERE jsonb_path_exists("ports", '$[*].ports[*].nickname ?(@ == "fff")')
1

同样支持NOT

2.3 时间搜索

SELECT * FROM "public"."tf_low_data_testUser" WHERE jsonb_path_exists("ports", '$[*].ports[*].date ?(@ >= "2022-01-02" && @ <= "2023-08-02")')
1

同样支持NOT

2.4 在列表

SELECT * FROM "public"."tf_low_data_testUser" WHERE jsonb_path_exists("ports", '$[*].ports[*].nickname ?(@ == "ggg" || @ == "fff")')
1

同样支持NOT

2.5 包含

和等值匹配一样

SELECT * FROM "public"."tf_low_data_testUser" WHERE jsonb_path_exists("ports", '$[*].ports[*].address ?(@ == "222")')
1

同样支持NOT