本篇文章讲解的主要内容是：日期操作函数(INTERVAL、EXTRACT)的使用以及如何确定一年是否为闰年及周的计算两个小案例
【SQL开发实战技巧】这一系列博主当作复习旧知识来进行写作，毕竟SQL开发在数据分析场景非常重要且基础，面试也会经常问SQL开发和调优经验，相信当我写完这一系列文章，也能再有所收获，未来面对SQL面试也能游刃有余~。

一、SYSDATE能得到的信息
经常看到有人因为不熟悉日期操作,获取相应信息的时候,要写很复杂的语句。下面举一个简单的例子。

SQL> select hiredate,
2 to_date(to_char(hiredate, 'yyyy-mm') || '-1', 'yyyy-mm-dd') as yuechu
3 from emp
4 where rownum <= 1;

HIREDATE YUECHU
----------- -----------
1980-12-17 1980-12-1

SQL>

其实要获取这个数据,只需要一个简单的函数就可以做到,而根本不需要多次转换：

SQL> SELECT hiredate AS 雇佣日期, trunc(hiredate, 'mm') AS 月初
2 from emp
3 where rownum <= 1;

雇佣日期 月初
----------- -----------
1980-12-17 1980-12-1

SQL>

下面列举几个常用的取值方式,希望对大家有用。

SQL> SELECT hiredate,
2 to_number(to_char(hiredate,'hh24')）时,
3 to_number(to_char(hiredate,'mi'))分,
4 to_number(to_char(hiredate,'ss')）秒,
5 to_number(to_char(hiredate,'dd'))日,
6 to_number(to_char(hiredate,'mm'))月,
7 to_number(to_char(hiredate,'yyyy'))年,
8 to_number(to_char(hiredate,'ddd'))年内第几天,
9 trunc(hiredate,'dd')一天之始,
10 trunc(hiredate,'day')周初,
11 trunc(hiredate,'dy')周初,
12 trunc(hiredate,'mm')月初,
13 last_day(hiredate)月未,
14 add_months(trunc(hiredate,'mm'),1)下月初,
15 trunc(hiredate,'yy')年初,
16 to_char(hiredate,'day')周几,
17 to_char(hiredate,'dy')周几,
18 to_char(hiredate,'month'）月份,
19 to_char(hiredate,'mm'）月份
20 FROM(SELECT hiredate+30/24/60/60+20/24/60+5/24 AS hiredate FROM emp WHERE ROWNUM<=1);

HIREDATE 时 分 秒 日 月 年 年内第几天 一天之始 周初 周初 月初 月未 下月初 年初 周几 周几 月份 月份
----------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------- ----------- ----------- ----------- ----------- ----------- ----------- --------------------------------------------------------------------------- --------------------------------------------------------------------------- --------------------------------------------------------------------------- ---------------------------------------------------------------------------
1980-12-17 5 20 30 17 12 1980 352 1980-12-17 1980-12-14 1980-12-14 1980-12-1 1980-12-31 1981-1-1 1980-1-1 星期三 星期三 12月 12

SQL>


需要注意的是上面last_day的用法,该函数返回的时分秒与参数中的一样,如果用该函数作为区间条件,会发生下面的情况。

SQL>
SQL> with t as (
2 select
3 to_date('1980-12-31 15:20:30','yyyy-mm-dd hh24:mi:ss') as d1,
4 to_date('1980-12-31 05:20:30','yyyy-mm-dd hh24:mi:ss') as d2
5 from dual
6 )
7 select d1,d2 from t where d1 between trunc(d2,'mm') and last_day(d2);

D1 D2
----------- -----------

SQL>

若要取一个月的数据,应该用下面的方式。

with t as (
select
to_date('1980-12-31 15:20:30','yyyy-mm-dd hh24:mi:ss') as d1,
to_date('1980-12-31 05:20:30','yyyy-mm-dd hh24:mi:ss') as d2
from dual
)
select d1,d2 from t where d1>=trunc(d2,'mm') and d1 D1 D2
----------------------------
1980-12-31 15:20:30 1980-12-31 5:20:30


二、INTERVAL
INTERVAL类型中保存的是时间间隔信息,可以通过对应的INTERVAL函数得到INTERVAL类型的数据。

SQL> select INTERVAL '2' year as "year",
2 INTERVAL '50' month as "month",
3 INTERVAL '99' day as "day",/*最大只能到99*/
4 INTERVAL'80' hour as "hour",
5 INTERVAL'90' minute as "minute",
6 INTERVAL'3.15' second as "second",
7 INTERVAL'2 12:30:59'DAY to second as "DAY to second",
8 INTERVAL'13-3' year to month as "Year to month" from dual;

year month day hour minute second DAY to second Year to month
------------------------------------------------------------------------------- ------------------------------------------------------------------------------- ------------------------------------------------------------------------------- ------------------------------------------------------------------------------- ------------------------------------------------------------------------------- ------------------------------------------------------------------------------- ------------------------------------------------------------------------------- -------------------------------------------------------------------------------
+02-00 +04-02 +99 00:00:00 +03 08:00:00 +00 01:30:00 +00 00:00:03.150000 +02 12:30:59.000000 +13-03

SQL>

当增加一个较复杂的时间段时,如上面的"02天12小时30分59秒“,通过INTERVAL实现显然更直观。

三、EXTRACT
与TO_CHAR一样,EXTRACT可以提取时间字段中的年、月、日、时、分、秒。不同的是,EXTRACT的返回值为NUMBER类型。

create table testa as
select
extract(year from systimestamp) as "year",
extract(month from systimestamp) as "month",
extract(day from systimestamp) as "day",
extract(hour from systimestamp) as "hour",
extract(minute from systimestamp) as "minute",
extract(second from systimestamp) as "second"
from dual;
SQL> desc testa;
Name Type Nullable Default Comments
------ ------ -------- ------- --------
year NUMBER Y
month NUMBER Y
day NUMBER Y
hour NUMBER Y
minute NUMBER Y
second NUMBER Y

SQL>


EXTRACT不能取DATE中的时、分、秒,示例如下：

SQL> select
2 extract(hour from sysdate) as "hour",
3 extract(minute from sysdate) as "minute",
4 extract(second from sysdate) as "second"
5 from dual;
select
extract(hour from sysdate) as "hour",
extract(minute from sysdate) as "minute",
extract(second from sysdate) as "second"
from dual

ORA-30076: 对析出来源无效的析出字段

SQL>

TO_CHAR可以,示例如下：

SQL> select
2 to_char(sysdate,'dd') as dd,
3 to_char(sysdate,'hh24') as hh,
4 to_char(sysdate,'mi') as mi,
5 to_char(sysdate,'ss') as ss
6 from dual;

DD HH MI SS
--------------------------------------------------------------------------- --------------------------------------------------------------------------- --------------------------------------------------------------------------- ---------------------------------------------------------------------------
15 14 43 30

SQL>

EXTRACT可以取INTERVAL中的信息,示例如下：

SQL> select extract(hour from it) as "hour"
2 from (
3 select INTERVAL'2 12:30:59'DAY to second as it from dual
4 );

hour
----------
12

SQL>

而TO_CHAR不行,示例如下：

SQL> select to_Char(it,'hh24') as "hour"
2 from (
3 select INTERVAL'2 12:30:59'DAY to second as it from dual
4 );

hour
--------------------
+02 12:30:59.000000

SQL>

四、确定一年是否为闰年
若要判断一年是否为闰年,只需要看二月的月末是哪一天就可以。

SQL> select trunc(sysdate, 'y') as 年初,
2 add_months(trunc(sysdate, 'y'), 1) as 二月初,
3 last_day(add_months(trunc(sysdate, 'y'), 1)) as 二月底,
4 to_char(last_day(add_months(trunc(sysdate, 'y'), 1)), 'dd') as 日
5 from dual;

年初 二月初 二月底 日
----------- ----------- ----------- ---------------------------------------------------------------------------
2023-1-1 2023-2-1 2023-2-28 28

SQL>

五、周的计算
SQL> with t as (
2 select trunc(sysdate,'YY')+(level-1) as 日期
3 from dual
4 connect by level<=8
5 )
6 select 日期,
7 /*返回值1代表周日,2代表周一*/
8 to_char(日期,'d') as d,
9 to_char(日期,'day') as day,
10 /*参数2中1代表周日,2代表周一*/
11 next_day(日期,1) as 下个周日,
12 /*ww的算法为每年1月1日为第一周开始,date+6为每一周结束*/
13 to_char(日期,'ww') as ww,
14 /*iw的算法为星期一至星期日算一周,且每年的第一个星期一为第一周*/
15 to_char(日期,'iw') as iw
16 from t;

日期 D DAY 下个周日 WW IW
----------- --------------------------------------------------------------------------- --------------------------------------------------------------------------- ----------- --------------------------------------------------------------------------- ---------------------------------------------------------------------------
2023-1-1 1 星期日 2023-1-8 01 52
2023-1-2 2 星期一 2023-1-8 01 01
2023-1-3 3 星期二 2023-1-8 01 01
2023-1-4 4 星期三 2023-1-8 01 01
2023-1-5 5 星期四 2023-1-8 01 01
2023-1-6 6 星期五 2023-1-8 01 01
2023-1-7 7 星期六 2023-1-8 01 01
2023-1-8 1 星期日 2023-1-15 02 01

8 rows selected


SQL>

注意以下两点：

参数"day"与字符集有关,所以提倡改用"d"。
WW与IW都是取“第几周”,只是两个参数的初始值不一样。
————————————————

免责申明：

本文系转载，版权归原作者所有，如若侵权请联系我们进行删除！

想了解或咨询更多有关袋鼠云大数据产品、行业解决方案、客户案例的朋友，浏览袋鼠云官网：https://www.dtstack.com/?src=bbs
同时，欢迎对大数据开源项目有兴趣的同学加入「袋鼠云开源框架钉钉技术群」，交流最新开源技术信息，群号码：30537511，项目地址：https://github.com/DTStack