MySQL statistical skills: on duplicate key update usage

ON DUPLICATE KEY UPDATE yes mysql Special grammar of , And INSERT INTO Use it together , It means that the record will be updated as soon as it exists , Otherwise add

INSERT INTO user(userid,username,age) 
VALUES(1,'ssy',20) ON DUPLICATE KEY UPDATE age = age + 1;

Perform result analysis :

• Suppose the record before this statement is not executed is like this ：

userid	username	age
1	ssy	20

Case one ：
If userid、username、age Any of the three fields is set PRIMARY KEY perhaps UNIQUE, Then the effect of this statement is UPDATE.
hypothesis userid It's set to PRIMARY KEY, So the one above SQL amount to ：

UPDATE user SET age = age + 1 WHERE userid = 1;

userid	username	age
1	ssy	21

It should be noted that ： In this case, only recognition is set to PRIMARY KEY Of userid Fields and by ON DUPLICATE KEY UPDATE Of age Field , Other fields in the statement are ignored .

The second case ：
If userid、username、age None of the three fields PRIMARY KEY perhaps UNIQUE, The effect of this sentence is INSERT INTO
above SQL Statement equivalent ：

INSERT INTO user(userid,username,age) VALUES(1,'ssy',20);

userid	username	age
1	ssy	20
2	ssy	20

Applicable scenario ： Statistics count

INSERT INTO … ON DUPLICATE KEY UPDATE … The function of is simply , Update when records exist , Otherwise insert . This is very suitable for quantity statistics .

For example, the user operation log statistics table is called count, Field is id,username,opnum,date

• id by PRIMARY KEY
• username and date Is not repeated UNIQUE Combine , That is, as long as username and date This pair of combinations cannot be repeated

CREATE TABLE `user`( 
	`id` INT UNSIGNED AUTO_INCREMENT PRIMARY KEY, 
	`username` VARCHAR(30) NOT NULL DEFAULT '', 
	`opnum` INT(11) NOT NULL DEFAULT '0', 
	`date` INT(11) NOT NULL DEFAULT '0', 
	UNIQUE KEY `unique` (`username`, `date`) 
)

The original data is ：

userid	username	opnum	date
1	ssy	0	20150424

If executed ：

INSERT INTO count(username,opnum,date) 
VALUES('ssy',5,'20150424') ON DUPLICATE KEY UPDATE opnum = opnum + 1;

The result is ：

userid	username	opnum	date
1	ssy	1	20150424

INSERT INTO count(username,opnum,date) VALUES('ssy',10,'20150425') 
ON DUPLICATE KEY UPDATE opnum = opnum + 1; 

INSERT INTO count(username,opnum,date) 
VALUES('dlm',10,'20150425') ON DUPLICATE KEY UPDATE opnum = opnum + 1;

The result is ：

userid	username	opnum	date
1	ssy	1	20150424
2	ssy	10	20150424
3	dlm	10	20150424

such , If the user ssy stay 2015 year 4 month 24 If there is any operation on this day , Just append the operand , If the user has no operation today , Then add a record of today's operation . That is, it realizes the daily statistics of multiple users .