MySQL5.7特性：JSON数据类型学习
概述 MySQL5.7的发行声明中，官方称之为里程碑式的版本，除了运行速度大幅度提升之外，还添加了之前版本没有的功能，如本文所述的原生JSON数据类型功能。在此版本之前，MySQL所有的JSON数据类型，全部是使用text等文本类型来实现的，数据的处理只能在应用代码级来实现，十分不方便。 什么是JSON类型 作为DBA，可能会对这个概念稍微有点陌生，但是对于开发者来说，这是一个十分熟悉的事物。 JSON(JavaScript Object Notation, JS 对象简谱) 是一种轻量级的数据交换格式。它基于 ECMAScript (欧洲计算机协会制定的js规范)的一个子集，采用完全独立于编程语言的文本格式来存储和表示数据。简洁和清晰的层次结构使得 JSON 成为理想的数据交换语言。 易于人阅读和编写，同时也易于机器解析和生成，并有效地提升网络传输效率。 MySQL原始JSON类型的优势在哪？ 原生的JSON优势如下： 存储上类似text,可以存非常大的数据。 存储在JSON列中的JSON文档的自动验证 。无效的文档会产生错误。 优化的存储格式。存储在JSON列中的JSON文档将 转换为内部格式，以允许对文档元素进行快速读取访问。 相比于传统形式，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中部分的数据进行索引。 MySQL的JSON类型 创建JSON类型表 创建一个基础的员工表，除了工号字段外，还有一个个人基础信息字段和一个个人能力信息字段 MySQL [test]> CREATE TABLE employee ( -> -> `empno` int(10) unsigned NOT NULL AUTO_INCREMENT, -> -> `basic_info` JSON NOT NULL, -> -> `skill_info` JSON NOT NULL, -> -> PRIMARY KEY (`empno`) -> -> ); Query OK, 0 rows affected (0.02 sec) 表的基础信息，其中JSON类型的字段，是不可以有默认值的，这点需要注意 MySQL [test]> desc employee; +------------+------------------+------+-----+---------+----------------+ | Field | Type | Null | Key | Default | Extra | +------------+------------------+------+-----+---------+----------------+ | empno | int(10) unsigned | NO | PRI | NULL | auto_increment | | basic_info | json | NO | | NULL | | | skill_info | json | NO | | NULL | | +------------+------------------+------+-----+---------+----------------+ 3 rows in set (0.00 sec) 试着插入几条数据 我们手动插入几条数据进这张表中，在前两条数据中，在个人能力信息上使用的是数组的方式，，后面两条则是使用对象的形式。在MySQL5.7.8版本后的JSON类型中，这两种都是可以的 INSERT INTO `employee` VALUES (1,'{"name": "wangyiyi", "age": "23" ,"from": "hangzhou"}', '["java", "go", "python"]'); INSERT INTO `employee` VALUES (2,'{"name": "linxue", "age": 24 ,"from": "shanghai"}', '["mysql", "oracle", "python"]'); INSERT INTO `employee` VALUES (3,'{"name": "zhaoqing", "age": 24 ,"from": "shanghai"}', '{"system": "linux","database": "mysql", "language": "python"}'); INSERT INTO `employee` VALUES (4,'{"name": "zhouxixi", "age": 30 ,"from": "nanjing"}', '{"system": ["linux","windows"],"database": ["mysql","oracle","postgresql"], "language": ["python","java","go"]}'); 插入多个数据后,表中内容为如下 MySQL [test]> select * from employee; +-------+-------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+ | empno | basic_info | skill_info | +-------+-------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+ | 1 | {"age": "23", "from": "hangzhou", "name": "wangyiyi"} | ["java", "go", "python"] | | 2 | {"age": 24, "from": "shanghai", "name": "linxue"} | ["mysql", "oracle", "python"] | | 3 | {"age": 24, "from": "shanghai", "name": "zhaoqing"} | {"system": "linux", "database": "mysql", "language": "python"} | | 4 | {"age": 30, "from": "nanjing", "name": "zhouxixi"} | {"system": ["linux", "windows"], "database": ["mysql", "oracle", "postgresql"], "language": ["python", "java", "go"]} | +-------+-------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+ 4 rows in set (0.00 sec) json数据查询方式 在插入了json类型的数据之后，可以针对JSON类型做一些特定的查询，如查询年龄大于20的记录在SQL的语句中使用 字段->.键名 就可以查询出所对应的键值 MySQL [test]> select * from employee WHERE basic_info->'$.age'> 20; +-------+-------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+ | empno | basic_info | skill_info | +-------+-------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+ | 1 | {"age": "23", "from": "hangzhou", "name": "wangyiyi"} | ["java", "go", "python"] | | 2 | {"age": 24, "from": "shanghai", "name": "linxue"} | ["mysql", "oracle", "python"] | | 3 | {"age": 28, "from": "shanghai", "name": "zhaoqing"} | {"system": "linux", "database": "mysql", "language": "go"} | | 4 | {"age": 30, "from": "nanjing", "name": "zhouxixi"} | {"system": ["linux", "windows"], "database": ["mysql", "oracle", "postgresql"], "language": ["python", "java", "go"]} | +-------+-------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+ 4 rows in set (0.00 sec) MySQL [test]> select * from employee WHERE basic_info->'$.age'< 20; Empty set (0.00 sec) 除了使用上述方式外，也可使用 提取json值的 函数 json_extract （使用函数的方式） MySQL [test]> select * from employee where json_extract(basic_info,'$.age') = 24; +-------+-----------------------------------------------------+----------------------------------------------------------------+ | empno | basic_info | skill_info | +-------+-----------------------------------------------------+----------------------------------------------------------------+ | 2 | {"age": 24, "from": "shanghai", "name": "linxue"} | ["mysql", "oracle", "python"] | | 3 | {"age": 24, "from": "shanghai", "name": "zhaoqing"} | {"system": "linux", "database": "mysql", "language": "python"} | +-------+-----------------------------------------------------+----------------------------------------------------------------+ 2 rows in set (0.00 sec) 对于数值查询也可做一个范围内查询，如下： MySQL [test]> select * from employee WHERE basic_info->'$.age' in (23, 24); +-------+-------------------------------------------------------+----------------------------------------------------------------+ | empno | basic_info | skill_info | +-------+-------------------------------------------------------+----------------------------------------------------------------+ | 1 | {"age": "23", "from": "hangzhou", "name": "wangyiyi"} | ["java", "go", "python"] | | 2 | {"age": 24, "from": "shanghai", "name": "linxue"} | ["mysql", "oracle", "python"] | | 3 | {"age": 24, "from": "shanghai", "name": "zhaoqing"} | {"system": "linux", "database": "mysql", "language": "python"} | +-------+-------------------------------------------------------+----------------------------------------------------------------+ 3 rows in set, 1 warning (0.00 sec) 因为 JSON 不同于字符串，所以如果用字符串和 JSON 字段比较，是不会相等的： 如下，直接使用字符串查询，查询不出来内容 MySQL [test]> select * from employee where basic_info = '{"age": 24, "from": "shanghai", "name": "linxue"}'; Empty set (0.00 sec) 可以通过 CAST 将字符串转成 JSON 的形式，如下： MySQL [test]> select * from employee where basic_info = CAST('{"age": 24, "from": "shanghai", "name": "linxue"}' AS JSON); +-------+---------------------------------------------------+-------------------------------+ | empno | basic_info | skill_info | +-------+---------------------------------------------------+-------------------------------+ | 2 | {"age": 24, "from": "shanghai", "name": "linxue"} | ["mysql", "oracle", "python"] | +-------+---------------------------------------------------+-------------------------------+ 1 row in set (0.00 sec) 查看单纯数组类型的函数JSON_CONTAINS MySQL [test]> select * from employee where JSON_CONTAINS (skill_info,'"mysql"'); +-------+---------------------------------------------------+-------------------------------+ | empno | basic_info | skill_info | +-------+---------------------------------------------------+-------------------------------+ | 2 | {"age": 24, "from": "shanghai", "name": "linxue"} | ["mysql", "oracle", "python"] | +-------+---------------------------------------------------+-------------------------------+ 1 row in set (0.00 sec) JSON_PRETTY函数： 以易于阅读的格式打印出JSON值便于在一些外部应用引用数据时，更方便的使用它 MySQL [test]> select JSON_PRETTY(basic_info) from employee; +---------------------------------------------------------------+ | JSON_PRETTY(basic_info) | +---------------------------------------------------------------+ | { "age": "23", "from": "hangzhou", "name": "wangyiyi" } | | { "age": 24, "from": "shanghai", "name": "linxue" } | | { "age": 28, "from": "shanghai", "name": "zhaoqing" } | | { "age": 30, "from": "nanjing", "name": "zhouxixi" } | +---------------------------------------------------------------+ 4 rows in set (0.00 sec) MySQL 5.7.22中添加了此功能，此函数返回用于存储JSON文档的二进制表示的字节数，用于查看当前JSON字段的存储大小 MySQL [test]> select skill_info,JSON_STORAGE_SIZE(skill_info) AS Size from employee; +-----------------------------------------------------------------------------------------------------------------------+------+ | skill_info | Size | +-----------------------------------------------------------------------------------------------------------------------+------+ | ["java", "go", "python"] | 29 | | ["mysql", "oracle", "python"] | 34 | | {"system": "linux", "database": "mysql", "language": "go"} | 63 | | {"system": ["linux", "windows"], "database": ["mysql", "oracle", "postgresql"], "language": ["python", "java", "go"]} | 137 | +-----------------------------------------------------------------------------------------------------------------------+------+ 4 rows in set (0.00 sec) 查询JSON字段的长度 MySQL [test]> select JSON_LENGTH(basic_info) from employee; +-------------------------+ | JSON_LENGTH(basic_info) | +-------------------------+ | 3 | | 3 | | 3 | | 3 | +-------------------------+ 4 rows in set (0.00 sec) 查看数据的类型：可以是对象，数组或标量类型 MySQL [test]> select JSON_TYPE(skill_info) from employee; +-----------------------+ | JSON_TYPE(skill_info) | +-----------------------+ | ARRAY | | ARRAY | | OBJECT | | OBJECT | +-----------------------+ 4 rows in set (0.00 sec) json数据修改方式 如果是整个 json 更新的话，和一般类型插入是一样的 json_array_insert是在指定下标插入，这是插入一般数组类型时的操作 MySQL [test]> SELECT json_array_insert(skill_info, '$[1]', 'php') from employee; +-----------------------------------------------------------------------------------------------------------------------+ | json_array_insert(skill_info, '$[1]', 'php') | +-----------------------------------------------------------------------------------------------------------------------+ | ["java", "php", "go", "python"] | | ["mysql", "php", "oracle", "python"] | | {"system": "linux", "database": "mysql", "language": "python"} | | {"system": ["linux", "windows"], "database": ["mysql", "oracle", "postgresql"], "language": ["python", "java", "go"]} | +-----------------------------------------------------------------------------------------------------------------------+ 4 rows in set (0.00 sec) 替换操作，也就是修改update操作，使用的是 json_replace 函数json_replace：只替换已经存在的旧值，不存在则忽略； MySQL [test]> update employee set skill_info = json_replace(skill_info, "$.language", "go") where empno = 3; Query OK, 1 row affected (0.00 sec) Rows matched: 1 Changed: 1 Warnings: 0 MySQL [test]> select * from employee where empno = 3; +-------+-----------------------------------------------------+------------------------------------------------------------+ | empno | basic_info | skill_info | +-------+-----------------------------------------------------+------------------------------------------------------------+ | 3 | {"age": 24, "from": "shanghai", "name": "zhaoqing"} | {"system": "linux", "database": "mysql", "language": "go"} | +-------+-----------------------------------------------------+------------------------------------------------------------+ 1 row in set (0.00 sec) json_set：替换旧值，并插入不存在的新值； MySQL [test]> update employee set basic_info = json_set(basic_info, "$.age", 28,"$.sex" ,"man") where empno = 3; Query OK, 1 row affected (0.01 sec) Rows matched: 1 Changed: 1 Warnings: 0 MySQL [test]> MySQL [test]> select * from employee where empno = 3; +-------+-------------------------------------------------------------------+------------------------------------------------------------+ | empno | basic_info | skill_info | +-------+-------------------------------------------------------------------+------------------------------------------------------------+ | 3 | {"age": 28, "sex": "man", "from": "shanghai", "name": "zhaoqing"} | {"system": "linux", "database": "mysql", "language": "go"} | +-------+-------------------------------------------------------------------+------------------------------------------------------------+ 1 row in set (0.00 sec) json_insert：插入新值，但不替换已经存在的旧值； MySQL [test]> update employee set basic_info = json_insert (basic_info, "$.age", 30, "$.phone" ,"123456789") where empno = 3; Query OK, 1 row affected (0.01 sec) Rows matched: 1 Changed: 1 Warnings: 0 MySQL [test]> select * from employee where empno = 3; +-------+-----------------------------------------------------------------------------------------+------------------------------------------------------------+ | empno | basic_info | skill_info | +-------+-----------------------------------------------------------------------------------------+------------------------------------------------------------+ | 3 | {"age": 28, "sex": "man", "from": "shanghai", "name": "zhaoqing", "phone": "123456789"} | {"system": "linux", "database": "mysql", "language": "go"} | +-------+-----------------------------------------------------------------------------------------+------------------------------------------------------------+ 1 row in set (0.00 sec) json_remove() 删除元素函数。 MySQL [test]> update employee set basic_info = json_remove (basic_info, "$.sex", "$.phone") where empno = 3; Query OK, 1 row affected (0.01 sec) Rows matched: 1 Changed: 1 Warnings: 0 MySQL [test]> MySQL [test]> select * from employee where empno = 3; +-------+-----------------------------------------------------+------------------------------------------------------------+ | empno | basic_info | skill_info | +-------+-----------------------------------------------------+------------------------------------------------------------+ | 3 | {"age": 28, "from": "shanghai", "name": "zhaoqing"} | {"system": "linux", "database": "mysql", "language": "go"} | +-------+-----------------------------------------------------+------------------------------------------------------------+ 1 row in set (0.00 sec) 结语 JSON数据类型是一个对开发十分友好的功能，有了它，MySQL的功能才更趋于完善。经常使用，会发现还有许多便捷的JSON函数能够在特定情况下帮到我们。详细信息也可以查看 https://dev.mysql.com/doc/refman/5.7/en/json-function-reference.html 云平-20190508
