Apply what you have learned, welcome to reprint, contact QQ:289325414 for more
Some of my own commonly used experience writing
table of Contents
Three, get the consecutive days of a week
Six, judge whether the field is empty
Seven, statistical SQL, add judgment in the function count(), sum(), count() de-duplication usage
Twelve, query whether there is data in the partition, and get the partition with data
1. Data condition filtering
select a from table where a=b field a do not use functions to convert
to_date, to_char, substr, will greatly slow down query efficiency
Second, the divisor is zero
The divisor is zero, and the error is avoided. Select decode(b,0,0,a/b) from dual;
Three, get the consecutive days of a week
SELECT to_char (to_date('2018-01-24')- LEVEL + 1, 'yyyy-mm-dd') today FROM
DUAL connect BY LEVEL <= 7;
Four, trunc
trunc function, trunc(sysdate,'DD') gets the yyyy-mm-dd standard date, removes the hour, minute and second, trunc(sysdate,'MM') gets a certain date on the first day of the month
Five, instr()
instr() function. Equivalent to like'%xxx%'
select * from test t
where instr(t.xxx,'xx')>0
or method usage instead
select * from test t
where instr(‘123，321’,t.xxx)>0
Six, judge whether the field is empty
select*from tab_i t where 1=nvl(t.col_x,1);
Not empty
select*from tab_i t where 1!=nvl(t.col_x,1);
Seven, statistical SQL, add judgment in the function count(), sum(), count() de-duplication usage
with t1 as (
select '张三' as name , '语文' as calls , '89' as scare from dual
union all
select '张三' as name , '语文' as calls , '88' as scare from dual
union all
select '张三' as name , '数学' as calls , null as scare from dual
union all
select '张三' as name , '英语' as calls , '68' as scare from dual
union all
select '李四' as name , '语文' as calls , '99' as scare from dual
union all
select '李四' as name , '数学' as calls , '66' as scare from dual
union all
select '李四' as name , '英语' as calls , null as scare from dual
)
select name,count(case when scare is not null then calls else null end) as There are several ways to write count for courses where the score is not null
,sum(case when scare is not null then 1 else 0 end) as a course with a non-empty score, there are several ways to write sum
, count(distinct case when scare is not null then calls else null end) as a course with a non-empty score There are several ways to go to reprint
from t1 group by name;
Eight, update method:
1. A data update B table data
update P_H_MME_QUALITY_CHECK A set A.LOWSINR0=
(SELECT B.LOWSINR0 FROM P_H_RST_XDR_COVERS B WHERE A.HOUR_ID=B.HOUR_ID
AND A.MSISDN=B.MSISDN AND A.VERSION_DATE=B.VERSION_DATE)
COMMIT;
Nine, user unlock
alter user username account unlock; (username refers to the database user name)
Ten, table unlock
SELECT 'alter system kill session ''' || c.sid || '' || ',' || c.serial# ||''';'
, a.object_id, a.session_id, b.object_name, c.*
FROM v$locked_object a, dba_objects b, v$session c
WHERE a.object_id = b.object_id
AND a.SESSION_ID = c.sid(+)
AND B.OBJECT_NAME = 'test_table'
ORDER BY logon_time;
11. Lookup table name
select table_name from user_tables where table_name like '%CJW%';
Twelve, query whether there is data in the partition, and get the partition with data
1. Analyze table test_table compute statistics for table for all indexes for all indexed columns; the
first step is usually after the partition data is stored in the database, call, call the analysis partition data, after analysis, the partition information of the table (USER_TAB_PARTITIONS) will be make it complete
2、SELECT substr(PARTITION_NAME,2) yyyymmdd FROM USER_TAB_PARTITIONS
WHERE TABLE_NAME = 'test_table'
AND SAMPLE_SIZE IS NOT NULL
The second step is real-time query, which can quickly figure out which partitions have data