你当前正在访问 Microsoft Azure Global Edition 技术文档网站。 如果需要访问由世纪互联运营的 Microsoft Azure 中国技术文档网站,请访问 https://docs.azure.cn。
GROUP BY (NoSQL 查询)
适用范围: NoSQL
GROUP BY
子句按照一个或多个指定属性的值来拆分查询的结果。
语法
<group_by_clause> ::= GROUP BY <scalar_expression_list>
<scalar_expression_list> ::=
<scalar_expression>
| <scalar_expression_list>, <scalar_expression>
参数
说明 | |
---|---|
<scalar_expression_list> |
指定用于对查询结果分组(或拆分)的表达式。 |
<scalar_expression> |
任何标量表达式都是允许的,但标量子查询和标量聚合除外。 每个标量表达式必须包含至少一个属性引用。 单个表达式的数目或每个表达式的多重性没有限制。 |
示例
本节中的示例使用了此参考项集。 每个项都包含一个 capabilities
对象,该对象可能包含 softwareDevelopment
和 mediaTrained
属性。
[
{
"name": "Jordan Mitchell",
"capabilities": {
"softwareDevelopment": "python",
"mediaTrained": true
},
"team": "Cloud software engineering"
},
{
"name": "Mikaela Lee",
"capabilities": {
"softwareDevelopment": "javascript",
"mediaTrained": false
},
"team": "Cloud software engineering"
},
{
"name": "Graham Barnes",
"capabilities": {
"softwareDevelopment": "c-sharp",
"mediaTrained": true
},
"team": "Cloud software engineering"
},
{
"name": "Hayden Cook",
"capabilities": {
"softwareDevelopment": "javascript",
"mediaTrained": true
},
"team": "Cloud software engineering"
},
{
"name": "Morgan Connors",
"capabilities": {
"mediaTrained": true
},
"team": "Cloud software engineering"
},
{
"name": "Devon Torres",
"capabilities": {
"softwareDevelopment": "python",
"mediaTrained": false
},
"team": "Cloud software engineering"
},
{
"name": "Sam Centrell",
"capabilities": {
"softwareDevelopment": "javascript",
"mediaTrained": true
},
"team": "Cloud software engineering"
}
]
在第一个示例中, GROUP BY
子句用于使用指定属性的值创建项组。
SELECT
e.capabilities.softwareDevelopment AS developmentLang
FROM
employees e
GROUP BY
e.capabilities.softwareDevelopment
[
{},
{
"developmentLang": "c-sharp"
},
{
"developmentLang": "javascript"
},
{
"developmentLang": "python"
}
]
在下一个示例中,(COUNT
) 的聚合系统函数与分组一起使用,以提供每个组的项总数。
SELECT
COUNT(1) AS trainedEmployees,
e.capabilities.softwareDevelopment AS developmentLang
FROM
employees e
GROUP BY
e.capabilities.softwareDevelopment
[
{
"trainedEmployees": 1
},
{
"trainedEmployees": 1,
"developmentLang": "c-sharp"
},
{
"trainedEmployees": 3,
"developmentLang": "javascript"
},
{
"trainedEmployees": 2,
"developmentLang": "python"
}
]
在此最后一个示例中,使用多个属性对项进行分组。
SELECT
COUNT(1) AS employeesWithThisTraining,
e.capabilities.softwareDevelopment AS developmentLang,
e.capabilities.mediaTrained AS mediaReady
FROM
employees e
GROUP BY
e.capabilities.softwareDevelopment,
e.capabilities.mediaTrained
[
{
"employeesWithThisTraining": 1,
"developmentLang": "python",
"mediaReady": true
},
{
"employeesWithThisTraining": 1,
"developmentLang": "javascript",
"mediaReady": false
},
{
"employeesWithThisTraining": 1,
"developmentLang": "c-sharp",
"mediaReady": true
},
{
"employeesWithThisTraining": 2,
"developmentLang": "javascript",
"mediaReady": true
},
{
"employeesWithThisTraining": 1,
"mediaReady": true
},
{
"employeesWithThisTraining": 1,
"developmentLang": "python",
"mediaReady": false
}
]
注解
- 当查询使用
GROUP BY
子句时,SELECT
子句只能包含包括在GROUP BY
子句中的属性和系统函数的一部分。 一个例外是聚合函数,此类函数可以出现在SELECT
子句中,但不需包含在GROUP BY
子句中。 也可始终在SELECT
子句中包含文本值。 GROUP BY
字句必须在SELECT
、FROM
和WHERE
字句之后,OFFSET LIMIT
字句之前。 不能将GROUP BY
与ORDER BY
子句一起使用。GROUP BY
子句不支持任何以下功能、属性或函数:- 别名属性或别名系统函数(在
SELECT
子句中,别名仍然是允许的) - 子查询
- 聚合系统函数(仅在
SELECT
子句中允许这些函数)
- 别名属性或别名系统函数(在
- 不支持查询包含使用
GROUP BY
的聚合系统函数和子查询。 - 跨分区
GROUP BY
查询最多可以有 21 个聚合系统函数。