如果将json格式数据存储到类似mysql这样的关系型数据库中,怎么查
MySQL 5.7.7 labs版本开始InnoDB存储引擎已经原生支持JSON格式,该格式不是简单的BLOB类似的替换。原生的JSON格式支持有以下的优势:
JSON数据有效性检查:BLOB类型无法在数据库层做这样的约束性检查
查询性能的提升:查询不需要遍历所有字符串才能找到数据
支持索引:通过虚拟列的功能可以对JSON中的部分数据进行索引
首先我们来看如何在MySQL中使用原生的JSON格式:
mysql> create table user ( uid int auto_increment,
-> data json,primary key(uid))engine=innodb;
Query OK, 0 rows affected (0.01 sec)
mysql> insert into user values (NULL,
-> '{"name":"David","mail":"jiangchengyao@gmail.com","address":"Shangahai"}');
Query OK, 1 row affected (0.00 sec)
mysql> insert into user values (NULL,'{"name":"Amy","mail":"amy@gmail.com"}');
Query OK, 1 row affected (0.00 sec)
可以看到我们新建了表user,并且将列data定义为了JSON类型。这意味着我们可以对插入的数据做JSON格式检查,确保其符合JSON格式的约束,如插入一条不合法的JSON数据会报如下错误:
mysql> insert into user values (NULL,"test");
ERROR 3130 (22032): Invalid JSON text: "Invalid value" at position 2 in value (or column) 'test'.
此外,正如前面所说的,MySQL 5.7提供了一系列函数来高效地处理JSON字符,而不是需要遍历所有字符来查找,这不得不说是对MariaDB dynamic column的巨大改进:
mysql> select jsn_extract(data, '$.name'),jsn_extract(data,'$.address') from user;
+-----------------------------+-------------------------------+
| jsn_extract(data, '$.name') | jsn_extract(data,'$.address') |
+-----------------------------+-------------------------------+
| "David" | "Shangahai" |
| "Amy" | NULL |
+-----------------------------+-------------------------------+
2 rows in set (0.00 sec)
当然,最令人的激动的功能应该是MySQL 5.7的虚拟列功能,通过传统的B+树索引即可实现对JSON格式部分属性的快速查询。使用方法是首先创建该虚拟列,然后在该虚拟列上创建索引:
mysql> ALTER TABLE user ADD user_name varchar(128)
-> GENERATED ALWAYS AS (jsn_extract(data,'$.name')) VIRTUAL;
Query OK, 0 rows affected (0.01 sec)
Records: 0 Duplicates: 0 Warnings: 0
mysql> select user_name from user;
+-----------+
| user_name |
+-----------+
| "Amy" |
| "David" |
+-----------+
2 rows in set (0.00 sec)
mysql> alter table user add index idx_username (user_name);
Query OK, 2 rows affected (0.01 sec)
Records: 2 Duplicates: 0 Warnings: 0
然后可以通过添加的索引对用户名进行快速的查询,这和普通类型的列查询一样。而通过explain可以验证优化器已经选择了在虚拟列上创建的新索引:
mysql> explain select * from user where user_name='"Amy"'\G
*************************** 1. row ***************************
id: 1
select_type: SIMPLE
table: user
partitions: NULL
type: ref
possible_keys: idx_username
key: idx_username
key_len: 131
ref: const
rows: 1
filtered: 100.00
Extra: NULL
1 row in set, 1 warning (0.00 sec)
可以发现MySQL 5.7对于JSON格式堪称完美,相信PostgreSQL阵营需要寻找新的策略来“攻击”MySQL了吧。如无意外,还是会停留在优化器这块,毕竟这块是目前MySQL必须要克服的最大问题,好在MySQL团队已经在重构优化器代码,相信更好的优化器将会在下一个版本中全面爆发。而一大堆文档数据库们已经哭晕在厕所了吧。
2020-04-20 · MySQL开源数据库领先者
我们知道,JSON是一种轻量级的数据交互的格式,大部分NO SQL数据库的存储都用JSON。MySQL从5.7开始支持JSON格式的数据存储,并且新增了很多JSON相关函数。MySQL 8.0 又带来了一个新的把JSON转换为TABLE的函数JSON_TABLE,实现了JSON到表的转换。
举例一
我们看下简单的例子:
简单定义一个两级JSON 对象
mysql> set @ytt='{"name":[{"a":"ytt","b":"action"}, {"a":"dble","b":"shard"},{"a":"mysql","b":"oracle"}]}';Query OK, 0 rows affected (0.00 sec)
第一级:
mysql> select json_keys(@ytt);+-----------------+| json_keys(@ytt) |+-----------------+| ["name"] |+-----------------+1 row in set (0.00 sec)
第二级:
mysql> select json_keys(@ytt,'$.name[0]');+-----------------------------+| json_keys(@ytt,'$.name[0]') |+-----------------------------+| ["a", "b"] |+-----------------------------+1 row in set (0.00 sec)
我们使用MySQL 8.0 的JSON_TABLE 来转换 @ytt。
mysql> select * from json_table(@ytt,'$.name[*]' columns (f1 varchar(10) path '$.a', f2 varchar(10) path '$.b')) as tt;
+-------+--------+
| f1 | f2 |
+-------+--------+
| ytt | action |
| dble | shard |
| mysql | oracle |
+-------+--------+
3 rows in set (0.00 sec)
set @json_str1 = ' { "query_block": { "select_id": 1, "cost_info": { "query_cost": "1.00" }, "table": { "table_name": "bigtable", "access_type": "const", "possible_keys": [ "id" ], "key": "id", "used_key_parts": [ "id" ], "key_length": "8", "ref": [ "const" ], "rows_examined_per_scan": 1, "rows_produced_per_join": 1, "filtered": "100.00", "cost_info": { "read_cost": "0.00", "eval_cost": "0.20", "prefix_cost": "0.00", "data_read_per_join": "176" }, "used_columns": [ "id", "log_time", "str1", "str2" ] } }}';
mysql> select json_keys(@json_str1) as 'first_object';+-----------------+| first_object |+-----------------+| ["query_block"] |+-----------------+1 row in set (0.00 sec)
mysql> select json_keys(@json_str1,'$.query_block') as 'second_object';+-------------------------------------+| second_object |+-------------------------------------+| ["table", "cost_info", "select_id"] |+-------------------------------------+1 row in set (0.00 sec)
mysql> select json_keys(@json_str1,'$.query_block.table') as 'third_object'\G*************************** 1. row ***************************third_object: ["key","ref","filtered","cost_info","key_length","table_name","access_type","used_columns","possible_keys","used_key_parts","rows_examined_per_scan","rows_produced_per_join"]1 row in set (0.01 sec)
mysql> select json_extract(@json_str1,'$.query_block.table.cost_info') as 'forth_object'\G*************************** 1. row ***************************forth_object: {"eval_cost":"0.20","read_cost":"0.00","prefix_cost":"0.00","data_read_per_join":"176"}1 row in set (0.00 sec)
SELECT * FROM JSON_TABLE(@json_str1,
"$.query_block"
COLUMNS(
rowid FOR ORDINALITY,
NESTED PATH '$.table'
COLUMNS (
a1_1 varchar(100) PATH '$.key',
a1_2 varchar(100) PATH '$.ref[0]',
a1_3 varchar(100) PATH '$.filtered',
nested path '$.cost_info'
columns (
a2_1 varchar(100) PATH '$.eval_cost' ,
a2_2 varchar(100) PATH '$.read_cost',
a2_3 varchar(100) PATH '$.prefix_cost',
a2_4 varchar(100) PATH '$.data_read_per_join'
),
a3 varchar(100) PATH '$.key_length',
a4 varchar(100) PATH '$.table_name',
a5 varchar(100) PATH '$.access_type',
a6 varchar(100) PATH '$.used_key_parts[0]',
a7 varchar(100) PATH '$.rows_examined_per_scan',
a8 varchar(100) PATH '$.rows_produced_per_join',
a9 varchar(100) PATH '$.key'
),
NESTED PATH '$.cost_info'
columns (
b1_1 varchar(100) path '$.query_cost'
),
c INT path "$.select_id"
)
) AS tt;
+-------+------+-------+--------+------+------+------+------+------+----------+-------+------+------+------+------+------+------+
| rowid | a1_1 | a1_2 | a1_3 | a2_1 | a2_2 | a2_3 | a2_4 | a3 | a4 | a5 | a6 | a7 | a8 | a9 | b1_1 | c |
+-------+------+-------+--------+------+------+------+------+------+----------+-------+------+------+------+------+------+------+
| 1 | id | const | 100.00 | 0.20 | 0.00 | 0.00 | 176 | 8 | bigtable | const | id | 1 | 1 | id | NULL | 1 |
| 1 | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | 1.00 | 1 |
+-------+------+-------+--------+------+------+------+------+------+----------+-------+------+------+------+------+------+------+
2 rows in set (0.00 sec)
举例二
再来一个复杂点的例子,用的是EXPLAIN 的JSON结果集。
JSON 串 @json_str1。
第一级:
第二级:
第三级:
第四级:
那我们把这个JSON 串转换为表。
当然,JSON_table 函数还有其他的用法,我这里不一一列举了,详细的参考手册。
请点击输入图片描述
修改回答