根雷尚一个程序员的点滴感悟2024-04-03T09:05:49.445Zhttps://www.genlei.cn/竣峰HexoMySQL with 语句的使用https://www.genlei.cn/2024/03/clujl4u5g004x88ln3r658ksw/2024-03-28T11:11:42.000Z2024-04-03T09:05:49.445Z插入
需求：给年龄小于18岁的用户加上学生角色，实现SQL如下
1 2 3 4
insert into s_user_role (user_id,role_id) with a as (select * from s_user u where u.age<18 ), b as (select * from s_role sr where sr.role_name='学生') select user_id,role_id from a,b
更新
需求：给管理员这个角色的用户年龄加1
1 2 3 4 5
with a as (select u.user_id,u.username,u.age from s_user u,s_user_role sur ,s_role sr where u.user_id=sur.user_id and sur.role_id=sr.role_id and sr.role_name='管理员' ) update s_user set age = age+1 where user_id in (select user_id from a)
with a as (select '2023-05-06' as dt, 12 as val union select '2023-05-10',110 union select '2023-05-08',80), b as (select max(dt) max_dt,min(dt) min_dt from a), z as (select DATE(DATE_SUB(CURDATE(),INTERVAL rn day)) as dt from (SELECT ROW_NUMBER() over() as rn FROM (SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) a ,(SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) b ,(SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) c ) t ), t as (select z.* from z,b where dt>=b.min_dt and dt <= b.max_dt ) select t.dt,ifnull(a.val,0) as val from t left join a on t.dt = a.dt order by t.dt;
看到上面的SQL 是不是有一种要放弃的感觉。太长了，确实太长了。我们来解析一下 都是干嘛的。我们这里用的是MySQL with 的语法，这个语法的好处就是可以把SQL写的贼长。然后还能年的懂。
先看：
1
with a as (select '2023-05-06' as dt, 12 as val union select '2023-05-10',110 union select '2023-05-08',80)
这一段就是我们造三行数据，实际业务中应该用真实的业务表数据替换。接下来
1
b as (select max(dt) max_dt,min(dt) min_dt from a),
这个就是找到业务里最大的日期，和最小的日期。以方便后面的补全。然后就是特别长的 z
1 2 3 4 5 6 7 8
z as (select DATE(DATE_SUB(CURDATE(),INTERVAL rn day)) as dt from (SELECT ROW_NUMBER() over() as rn FROM (SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) a ,(SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) b ,(SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) c ) t ),
create view vn as SELECT ROW_NUMBER() over() as rn FROM (SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) a ,(SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) b ,(SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) c
这样一个 1 到 1000 的视图就创建好了。前面的SQL 也可以简化为：
1 2 3 4 5
with a as (select '2023-05-06' as dt, 12 as val union select '2023-05-10',110 union select '2023-05-08',80), b as (select max(dt) max_dt,min(dt) min_dt from a), z as (select DATE(DATE_SUB(CURDATE(),INTERVAL rn day)) as dt from vn), t as (select z.* from z,b where dt>=b.min_dt and dt <= b.max_dt ) select t.dt,ifnull(a.val,0) from t left join a on t.dt = a.dt order by t.dt;
with a as (select '2023-04' as dt, 12 as val union select '2023-06',110 union select '2023-07',80), b as (select max(dt) max_dt,min(dt) min_dt from a), z as (select date_format(DATE_SUB(CURDATE(),INTERVAL rn month),'%Y-%m') as dt from vn), t as (select z.* from z,b where dt>=b.min_dt and dt <= b.max_dt ) select t.dt,ifnull(a.val,0) from t left join a on t.dt = a.dt order by t.dt;
]]><h2 id="日期补全"><a href="#日期补全" class="headerlink" title="日期补全"></a>日期补全</h2><p>在数据统计时，有时候会出现某天没有数据的情况，比如某天某类产品销量为0，这样就没有真实订单。我们统计的时候就可能出现下面的数SQL Server 事务引起的表锁问题https://www.genlei.cn/2023/07/sql-server-is_read_committed_snapshot_on/2023-07-04T04:19:33.000Z2024-04-03T09:05:49.445Z最近在查一个SQL Server 事务引起的表锁问题，折腾好久才发现要开启 快照读取模式。
1 2 3 4 5 6
-- 查询数据库是否开启了 快照读取模式 select name, is_read_committed_snapshot_on from sys.databases
-- 打开 快照读取模式 ALTER DATABASE my_db SET READ_COMMITTED_SNAPSHOT ON
]]><p>最近在查一个SQL Server 事务引起的表锁问题，折腾好久才发现要开启 快照读取模式。</p>
<figure class="highlight plaintext"><table><tr><td class="gutter"><pre><span class="linMySQL查询最近1000天的日期https://www.genlei.cn/2023/06/mysql-latest-1000/2023-06-27T13:19:33.000Z2024-04-03T09:05:49.445Z直接看SQL
1 2 3 4 5 6 7 8
select DATE(DATE_SUB(CURDATE(),INTERVAL rn day)) as date from (SELECT ROW_NUMBER() over() as rn FROM (SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) a ,(SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) b ,(SELECT 0 AS N UNION SELECT 1 UNION SELECT 2 UNION SELECT 3 UNION SELECT 4 UNION SELECT 5 UNION SELECT 6 UNION SELECT 7 UNION SELECT 8 UNION SELECT 9) c ) t
select `month`,profit, lag(profit) OVER w AS 上月, (profit - lag(profit) OVER w) as 环比, 100*(profit - lag(profit) OVER w)/lag(profit) OVER w as 环比比例 from my_stat WINDOW w AS (ORDER BY `month`);
partition_clause 分区定义, 就是定义查询结果如何分组，有点类似group by 的意思。
order_clause 排序定义,定义窗口内容排序字段及方式，如前面示例里的 order by `month` 。如果省略了order by , 那就按查询到结果的顺序来排序。
frame_clause frame_clause(框架子句)指定如何定义子集
同比环比计算
还是先看最终SQL
1 2 3 4 5 6 7 8 9 10 11
select `month`,profit, lag(profit) OVER w AS 上月, (profit - lag(profit) OVER w) as 环比, 100*(profit - lag(profit) OVER w)/lag(profit) OVER w as 环比比例, lag(profit,12) OVER w AS 上年同月, (profit - lag(profit,12) OVER w) as 同比, 100*(profit - lag(profit,12) OVER w)/lag(profit,12) OVER w as 同比比例 from my_stat WINDOW w AS (ORDER BY `month`);
String str = "[{\"name\":\"tom\",\"age\":16},{\"name\":\"lucy\",\"age\":15}]"; List<Student> list = objectMapper.readValue(str, new TypeReference<List<Student>>(){});
把json文件解析为Java 数组
1 2 3 4
String str = "[{\"name\":\"tom\",\"age\":16},{\"name\":\"lucy\",\"age\":15}]"; ObjectMapper objectMapper = new ObjectMapper(); objectMapper.configure(DeserializationFeature.USE_JAVA_ARRAY_FOR_JSON_ARRAY, true); Student[] students = objectMapper.readValue(str, Student[].class);
把json文本解析为Java Map
借助 TypeReference
1 2
String str = "{\"name\":\"tom\",\"age\":16}"; Map<String,Object> map = objectMapper.readValue(str, new TypeReference<Map<String,Object>>(){});
public static void main(String[] args){ Stream.of("a","b","c","","e","f").takeWhile(s->!s.isEmpty()) .forEach(System.out::print); System.out.println("\n==="); Stream.of("a","b","c","","e","f").dropWhile(s-> !s.isEmpty()) .forEach(System.out::print); System.out.println("\n==="); IntStream.iterate(3, x -> x < 10, x -> x+ 3).forEach(System.out::println); System.out.println("==="); List list = Stream.ofNullable(null).collect(Collectors.toList()); System.out.println(list.size()); }
public void ifPresentOrElse(Consumer<? super T> action, Runnable emptyAction)
示例代码：
1 2 3 4 5 6 7
public static void main(String[] args){ Optional<String> op = Optional.of("A"); op.ifPresentOrElse(t->System.out.println(t),()->System.out.println("Empty"));
public class RecordTest { public static void main(String[] args){ Point p = new Point(1,2); System.out.println(p.x()); System.out.println(p.y()); System.out.println(p.toString()); } }
它适用于，简单不可变对象的封装。在某些场景下，可以替代lombok
19. sealed class 密封类
先看代码
1 2 3 4 5 6
public sealed class Person permits Female, Male { } public final class Male extends Person{ } public final class Female extends Person { }
上面我们定义了三个类，Person, Male 和 Female，即 人，男人，女人。 密封类就是增加了一种可能，明确定义某个类只能被指定的类继承。其它类无法对其进行继承。
从上面的例子我们看到 Person类前面使用了 sealed class 对其进行声明，在后面通过 permits 关键字指定了它只能被Female, Male这两个类进行继承。
Female, Male 这两个类在声明的时候必须指定它是final,还是 sealed ，还是non-sealed。如果指定为non-sealed,那么它就以可以被随意继承了。
看下面的代码：
1 2 3 4 5 6 7 8 9 10
public sealed class Person permits Female, Male { } public final class Female extends Person { } public sealed class Male extends Person permits OldMale { } public non-sealed class OldMale extends Male{ } public class CityOldMale extends OldMale{ }
]]><p><strong>声明：本人从未参与任何健康码系统的开发，此文是我基于自己对健康码的使用，相关新闻的阅读，按我自己的理解，整理出来的文章。如果跟某些系统有技术上的相似，纯属巧合。</strong></p>
<p><strong>摘要：</strong> 本文尝试从一个旁观者的sql server 读取表结构https://www.genlei.cn/2022/09/207/2022-09-27T05:26:18.000Z2024-04-03T09:05:49.451Z
SELECT col.name AS name, t.name AS type, Cast(col.length AS VARCHAR(80)) AS length, Cast(col.xprec AS VARCHAR(80)) AS precision, CASE WHEN EXISTS (SELECT 1 FROM dbo.sysindexes si INNER JOIN dbo.sysindexkeys sik ON si.id = sik.id AND si.indid = sik.indid INNER JOIN dbo.syscolumns sc ON sc.id = sik.id AND sc.colid = sik.colid INNER JOIN dbo.sysobjects so ON so.name = si.name AND so.xtype = 'PK' WHERE sc.id = col.id AND sc.colid = col.colid) THEN 'YES' ELSE 'NO' END AS 'key', CASE WHEN col.isnullable = 1 THEN 'NO' ELSE 'YES' END AS isnullable, Isnull(ep.[value], NULL) AS comment FROM dbo.syscolumns col LEFT JOIN dbo.systypes t ON col.xtype = t.xusertype LEFT JOIN sys.extended_properties ep ON col.id = ep.major_id AND col.colid = ep.minor_id INNER JOIN dbo.sysobjects obj ON col.id = obj.id AND obj.status >= 0 LEFT JOIN sys.schemas s ON obj.uid = s.schema_id WHERE s.name = 'schema_name' AND obj.name = ( 'table_name' );
终端层 终端层的说明请参考本文 一、主要功能中的 接入终端 接入层 接入层是指网络及接入中间件，终端通过接入层访问服务层。