有奖捉虫:云通信与企业服务文档专题,速来> HOT
TDSQL PostgreSQL版 不只是一个分布式关系型数据库系统,同时它还支持非关系数据类型 json。json 数据类型用来存储 JSON(JavaScript Object Notation)数据。这种数据也可以被存储为 text,但是 json 数据类型的优势在于能强制要求每个被存储的值符合 json 规则。 也有很多 json 相关的函数和操作符可以用于存储在这些数据类型中的数据。
json 数据类型有 json 和 jsonb,它们接受完全相同的值集合作为输入,主要的区别是效率。json 数据类型存储输入文本的精准拷贝,处理函数必须在每次执行时必须重新解析该数据。而 jsonb 数据被存储在一种分解好的二进制格式中,它在输入时要稍慢一些,因为需要做附加的转换。但是 jsonb 在处理时要快很多,因为不需要解析。jsonb 也支持索引,这也是其优势。

json 应用

创建 json 类型字段表

postgres=# create table t_json(id int,f_json json);
NOTICE: Replica identity is needed for shard table, please add to this table through "alter table" command.
CREATE TABLE

插入数据

postgres=# insert into t_json values(1,'{"col1":1,"col2":"tdsql_pg"}');
INSERT 0 1
postgres=# insert into t_json values(2,'{"col1":1,"col2":"tdsql_pg","col3":"pgxz"}');
INSERT 0 1
postgres=# select * from t_json;
id | f_json
----+-----------------------------------------
1 | {"col1":1,"col2":"tdsql_pg"}
2 | {"col1":1,"col2":"tdsql_pg","col3":"pgxz"}
(2 rows)

通过键获得 json 对象域

postgres=# select f_json ->'col2' as col2 ,f_json -> 'col3' as col3 from t_json;
col2 | col3
---------+--------
"tdsql_pg" |
"tdsql_pg" | "pgxz"
(2 rows)

以文本形式获取对象值

postgres=# select f_json ->>'col2' as col2 ,f_json ->> 'col3' as col3 from t_json;
col2 | col3
-------+------
tdsql_pg |
tdsql_pg | pgxz
(2 rows)

postgres=# select f_json ->>'col2' as col2 ,f_json ->> 'col3' as col3 from t_json where f_json ->> 'col3' is not null;
col2 | col3
-------+------
tdsql_pg | pgxz
(1 row)

jsonb 应用

创建 jsonb 类型字段表

postgres=# create table t_jsonb(id int,f_jsonb jsonb);
NOTICE: Replica identity is needed for shard table, please add to this table through "alter table" command.
CREATE TABLE
postgres=#

插入数据

postgres=# insert into t_jsonb values(1,'{"col1":1,"col2":"tdsql_pg"}');
INSERT 0 1
postgres=# insert into t_jsonb values(2,'{"col1":1,"col2":"tdsql_pg","col3":"pgxz"}');
INSERT 0 1

postgres=# select * from t_jsonb;
id | f_jsonb
----+----------------------------------------------
1 | {"col1": 1, "col2": "tdsql_pg"}
2 | {"col1": 1, "col2": "tdsql_pg", "col3": "pgxz"}
(2 rows)
jsonb 插入时会移除重复的键,如下所示。
postgres=# insert into t_jsonb values(3,'{"col1":1,"col2":"tdsql_pg","col2":"pgxz"}');
INSERT 0 1
postgres=# select * from t_jsonb;
id | f_jsonb
----+----------------------------------------------
1 | {"col1": 1, "col2": "tdsql_pg"}
3 | {"col1": 1, "col2": "pgxz"}
2 | {"col1": 1, "col2": "tdsql_pg", "col3": "pgxz"}
(3 rows)

更新数据

增加元素。
postgres=# update t_jsonb set f_jsonb = f_jsonb || '{"col3":"pgxz"}'::jsonb where id=1;
UPDATE 1
更新原来的元素。
postgres=# update t_jsonb set f_jsonb = f_jsonb || '{"col2":"tdsql_pg"}'::jsonb where id=3;
UPDATE 1

postgres=# select * from t_jsonb;
id | f_jsonb
----+----------------------------------------------
2 | {"col1": 1, "col2": "tdsql_pg", "col3": "pgxz"}
1 | {"col1": 1, "col2": "tdsql_pg", "col3": "pgxz"}
3 | {"col1": 1, "col2": "tdsql_pg"}
(3 rows)
删除某个键。
postgres=# update t_jsonb set f_jsonb = f_jsonb - 'col3';
UPDATE 3

postgres=# select * from t_jsonb;
id | f_jsonb
----+------------------------------
2 | {"col1": 1, "col2": "tdsql_pg"}
1 | {"col1": 1, "col2": "tdsql_pg"}
3 | {"col1": 1, "col2": "tdsql_pg"}
(3 rows)

jsonb_set() 函数更新数据

jsonb_set(target jsonb, path text[], new_value jsonb, [create_missing boolean])
说明
target 指要更新的数据源,path 指路径,new_value 指更新后的键值,create_missing 值为 true 表示如果键不存在则添加,create_missing 值为 false 表示如果键不存在则不添加。
postgres=# update t_jsonb set f_jsonb = jsonb_set( f_jsonb , '{col}' , '"pgxz"' , true ) where id=1;
UPDATE 1
postgres=# update t_jsonb set f_jsonb = jsonb_set( f_jsonb , '{col}' , '"pgxz"' , false ) where id=2;
UPDATE 1
postgres=# update t_jsonb set f_jsonb = jsonb_set( f_jsonb , '{col2}' , '"pgxz"' , false ) where id=3;
UPDATE 1
postgres=# select * from t_jsonb;
id | f_jsonb
----+---------------------------------------------
1 | {"col": "pgxz", "col1": 1, "col2": "tdsql_pg"}
2 | {"col1": 1, "col2": "tdsql_pg"}
3 | {"col1": 1, "col2": "pgxz"}
(3 rows)

jsonb 函数应用

jsonb_each() 将 json 对象转变键和值

postgres=# select f_jsonb from t_jsonb where id=1;
f_jsonb
---------------------------------------------
{"col": "pgxz", "col1": 1, "col2": "tdsql_pg"}
(1 row)

postgres=# select * from jsonb_each((select f_jsonb from t_jsonb where id=1));
key | value
------+---------
col | "pgxz"
col1 | 1
col2 | "tdsql_pg"
(3 rows)

jsonb_each_text() 将 json 对象转变文本类型的键和值

postgres=# select * from jsonb_each_text((select f_jsonb from t_jsonb where id=1));
key | value
------+-------
col | pgxz
col1 | 1
col2 | tdsql_pg
(3 rows)

row_to_json() 将一行记录变成一个 json 对象

postgres=# \\d+ tdsql_pg
Table "public.tdsql_pg"
Column | Type | Collation | Nullable | Default | Storage | Stats target | Description
----------+---------+-----------+----------+---------+----------+--------------+-------------
id | integer | | not null | | plain | |
nickname | text | | | | extended | |
Indexes:
"tdsql_pg_pkey" PRIMARY KEY, btree (id)
Distribute By: SHARD(id)
Location Nodes: ALL DATANODES

postgres=# select * from tdsql_pg;
id | nickname
----+----------
1 | tdsql_pg
2 | pgxz
(2 rows)

postgres=# select row_to_json(tdsql_pg) from tdsql_pg;
row_to_json
-----------------------------
{"id":1,"nickname":"tdsql_pg"}
{"id":2,"nickname":"pgxz"}
(2 rows)

json_object_keys() 返回一个对象中所有的键

postgres=# select * from json_object_keys((select f_jsonb from t_jsonb where id=1)::json);
json_object_keys
------------------
col
col1
col2
(3 rows)

jsonb 索引使用

tdsql_pg 为文档 jsonb 提供了 GIN 索引,GIN 索引可以被用来有效地搜索在大量 jsonb 文档(数据)中出现的键或者键值对。

创建 jsonb 索引

postgres=# create index t_jsonb_f_jsonb_idx on t_jsonb using gin(f_jsonb);
CREATE INDEX

postgres=# \\d+ t_jsonb
Table "public.t_jsonb"
Column | Type | Collation | Nullable | Default | Storage | Stats target | Description
---------+---------+-----------+----------+---------+----------+--------------+-------------
id | integer | | | | plain | |
f_jsonb | jsonb | | | | extended | |
Indexes:
"t_jsonb_f_jsonb_idx" gin (f_jsonb)
Distribute By: SHARD(id)
Location Nodes: ALL DATANODES

测试查询的性能

postgres=# select count(1) from t_jsonb;
count
----------
10000000
(1 row)

postgres=# analyze t_jsonb;
ANALYZE
没有索引开销。
postgres=# select * from t_jsonb where f_jsonb @> '{"col1":9999}';
id | f_jsonb
------+--------------------------------
9999 | {"col1": 9999, "col2": "9999"}
9999 | {"col1": 9999, "col2": "9999"}
9999 | {"col1": 9999, "col2": "9999"}
9999 | {"col1": 9999, "col2": "9999"}
9999 | {"col1": 9999, "col2": "9999"}
(5 rows)
Time: 2473.488 ms (00:02.473)
有索引开销。
postgres=# select * from t_jsonb where f_jsonb @> '{"col1":9999}';
id | f_jsonb
------+--------------------------------
9999 | {"col1": 9999, "col2": "9999"}
9999 | {"col1": 9999, "col2": "9999"}
9999 | {"col1": 9999, "col2": "9999"}
9999 | {"col1": 9999, "col2": "9999"}
9999 | {"col1": 9999, "col2": "9999"}
(5 rows)

Time: 217.968 ms