So, I've got a table roughly as follows:
CREATE TABLE CUSTOMER (
CUSTID INT NOT NULL AUTO_INCREMENT,
NAME CHAR (45),
CONSTRAINT CUSTOMER_PRIMARY_KEY PRIMARY KEY (CUSTID))
AUTO_INCREMENT = 100;
I'm auto incrementing the CUSTID so that it's possible to simply insert a name and have it created with the next available CUSTID. However, I also want to ensure that it isn't possible to set the CUSTID value to zero, either on creation of the row or on update so I've constructed the following trigger:
DELIMITER $$
CREATE TRIGGER `custid_before_insert` BEFORE INSERT ON `CUSTOMER`
FOR EACH ROW
BEGIN
IF (NEW.CUSTID) <= 0 THEN
SIGNAL SQLSTATE '12345'
SET MESSAGE_TEXT = 'Check constraint on CUSTOMER.CUSTID failed';
END IF;
END$$
CREATE TRIGGER `custid_before_update` BEFORE UPDATE ON `CUSTOMER`
FOR EACH ROW
BEGIN
IF (NEW.CUSTID) <= 0 THEN
SIGNAL SQLSTATE '12345'
SET MESSAGE_TEXT = 'Check constraint on CUSTOMER.CUSTID failed';
END IF;
END$$
DELIMITER ;
Unfortunately in my blissful ignorance of how AUTO_INCREMENT worked, I've come to the conclusion that this is the wrong way to go about this. Trying to insert a customer with no CUSTID value is tripping the trigger causing the insert to fail which I presume is due to the value being a zero before insertion when AUTO_INCREMENT assigns it a value.
Would the best way to do this really be to change the trigger to occur after the insert and delete the row or is there a better way to do this to just throw an error?
The insert trigger is not needed.
From Auto_Increment
No value was specified for the AUTO_INCREMENT column, so MySQL
assigned sequence numbers automatically. You can also explicitly
assign 0 to the column to generate sequence numbers.
E.G.
create table t(id int auto_increment, primary key(id));
insert into t(id) values (0);
select id from t;
# 1
Update:
To allow the insert to complete when CUSTID is not specified,
INSERT INTO customer(name) VALUES('Chuck');
check for null in the trigger:
IF NEW.CUSTID IS NOT NULL AND NEW.CUSTID <= 0 THEN
Inserting '0' into an auto-increment column causes it to increment the same as inserting NULL, so you really neither need nor want the INSERT trigger. Try it with just the UPDATE trigger.
Related
doing my assignment on Databases. Working with MySQL, my task is to create database with several tables. One of them is the table of workers in which primary key is WId, which is the String of 16 A-Z and 0-9 elements. I need to make a constraint such that it is not possible to put there !##$%^ etc.
create table Worker ( WId char(16) primary key,
Name char(10),
Surname char(20),...);
Thanks.
You can always use the SIGNAL keyword to return an error:
EDIT 1:
Based on #Gordon_linoff comment: you would have to do both BEFORE INSERT and BEFORE UPDATE in case if you expect the Id to change.
CREATE TRIGGER myTrigger
BEFORE INSERT
ON worker
FOR EACH ROW
BEGIN
IF NOT (NEW.WId REGEXP '^[A-Za-z0-9]+$')
THEN
BEGIN
SIGNAL sqlstate '45000' set message_text = 'TADA!';
END;
END IF;
END;
I have a table that has two columns to store id from another table. Column1 gets id from ABC table and Column2 also gets id from that table but letter is called parent ID, so with this information I know who is parent of who.
Now I want to create a constraint not to ever let both columns to get same id.
The following did not work:
ALTER TABLE id_parent_table
ADD CHECK (parent_id != main_id)
This is still allowing to insert two identical numbers.
This is now supported as of MySQL 8.0.16.
See https://dev.mysql.com/doc/refman/8.0/en/create-table-check-constraints.html
mysql> create table id_parent_table (
-> main_id bigint unsigned not null,
-> parent_id bigint unsigned not null,
-> constraint columns_cannot_equal check (main_id <> parent_id)
-> );
Query OK, 0 rows affected (0.38 sec)
mysql> insert into id_parent_table (main_id, parent_id) values (1, 1);
ERROR 3819 (HY000): Check constraint 'columns_cannot_equal' is violated.
Apparently, MySQL does not support check constraints. To quote the online reference:
The CHECK clause is parsed but ignored by all storage engines.
You could, alternatively, use a trigger to fail such an insert or update:
EDIT: MySQL doesn't support a single trigger on two events, so you'd have to have two different triggers:
delimiter //
CREATE TRIGGER id_parent_table_check_insert_trg
BEFORE INSERT ON id_parent_table
FOR EACH ROW
BEGIN
DECLARE msg varchar(255);
IF new.parent_id = new.main_id THEN
SET msg = 'parent_id and main_id should be different';
SIGNAL SQLSTATE '45000' SET message_text = msg;
END IF;
END
//
CREATE TRIGGER id_parent_table_check_update_trg
BEFORE UPDATE ON id_parent_table
FOR EACH ROW
BEGIN
DECLARE msg varchar(255);
IF new.parent_id = new.main_id THEN
SET msg = 'parent_id and main_id should be different';
SIGNAL SQLSTATE '45000' SET message_text = msg;
END IF;
END
//
I have a problem with the following sql code executed in MySQL Workbench 5.6. The table is orderly created, the trigger creation code is also run without errors. But when I try to run the both insert into statements I have an error Error Code: 1054. Unknown column 'Pass' in 'field list'. If I run the last select statements it is also runned as expected!?
I really can not imagine what is happening after running the trigger creation script!?
If I do not run the triger creation code before the insert into statements they are executed as expected and the data is inserted in the columns.
/*TASK 15: Write a SQL statement to create a table Users. Users should have username,
password, full name and last login time. Choose appropriate data types for the table fields.
Define a primary key column with a primary key constraint. Define the primary key column
as identity to facilitate inserting records. Define unique constraint to avoid repeating
usernames. Define a check constraint to ensure the password is at least 5 characters long.
*/
create table Users(
UserID int auto_increment not null,
Username nvarchar(50) not null,
Pass varchar(100) not null,
FullName nvarchar(100),
LastLoginTime datetime,
constraint PK_Users primary key(UserID),
constraint UK_Users_Username UNIQUE(Username),
constraint CH_Users_Password CHECK(length(users.Pass)>=5)
);
DELIMITER $$
CREATE TRIGGER `TR_BeforeInsert_Users`
BEFORE INSERT ON `users`
FOR EACH ROW
BEGIN
IF LENGTH(`Pass` ) < 5 THEN
SIGNAL SQLSTATE '12345'
SET MESSAGE_TEXT = 'check constraint on Pass failed';
END IF;
END$$
delimiter ;
Insert into Users(Username, Pass, FullName,lastlogintime) values('ttitto','alabala','Todor', now());
Insert into Users(Username, Pass, FullName,lastlogintime) values('ttitt','ala','Todor', now());
select length(Pass) from users;
Within triggers, one must reference columns with either the NEW or OLD pseudo-table prefix. In your case:
DELIMITER $$
CREATE TRIGGER `TR_BeforeInsert_Users`
BEFORE INSERT ON `users`
FOR EACH ROW
BEGIN
IF CHAR_LENGTH(NEW.Pass) < 5 THEN
SIGNAL SQLSTATE '12345'
SET MESSAGE_TEXT = 'check constraint on Pass failed';
END IF;
END$$
delimiter ;
I have the following table.
CREATE TABLE people(
first_name VARCHAR(128) NOT NULL,
nick_name VARCHAR(128) NULL
)
I would like to prevent people from having their nickname be the same as their firstname if they attempt that insertion. I do not want to create an index on either of the columns just a rule to prevent the insertion of records where the first_name and nick_name are the same.
Is there a way to create a rule to prevent insertion of records where the first_name would equal the nick_name?
CREATE TRIGGER `nicknameCheck` BEFORE INSERT ON `people` FOR EACH ROW begin
IF (new.first_name = new.nick_name) THEN
SET new.nick_name = null;
END IF;
END
Or you can set first_name to NULL which will cause SQL error and you can handle it and show some warning.
You only need triggers for BEFORE INSERT and BEFORE UPDATE. Let these check the values and abort the operation, if they are equal.
Caveat: On older but still widely used versions of MySQL (before 5.5 IIRC) you need to do something bad, such as read from the written table or easier read from an inexistant table/column (in order to abort).
AFTER INSERT trigger to test and remove if same ...
CREATE TABLE ek_test (
id INT PRIMARY KEY NOT NULL AUTO_INCREMENT,
one INT NOT NULL,
two INT NOT NULL
);
delimiter //
CREATE TRIGGER ek_test_one_two_differ AFTER INSERT ON ek_test
FOR EACH ROW
BEGIN
IF (new.one = new.two) THEN
DELETE FROM ek_test WHERE id = new.id;
END IF;
END//
delimiter ;
INSERT INTO ek_test (one, two) VALUES (1, 1);
SELECT * FROM ek_test;
NOTE you will also need AFTER UPDATE trigger.
I have a table, which includes an ENUM field. I would like for a particular value of ENUM to require uniqueness with the reference ID column. So let's say I have this:
`id` INT UNSIGNED NOT NULL AUTO_INCREMENT,
`type` ENUM('single','multi') NOT NULL,
`refid` INT UNSIGNED NOT NULL,
`extra` TEXT,
PRIMARY KEY (`id`)
I want (type,refid) to be a unique key, but only if type is single. This can't be done with conventional keys, so I figured I would use a Trigger to detect insertion of a row, check if type='single', look for a row with type='single' and refid=new.refid, and throw a duplicate entry error if one is found.
I'm using MySQL 5.5, so SIGNAL SQLSTATE is available to me. Can I use this to fire a Duplicate Key error in order to process the ON DUPLICATE KEY UPDATE part of the query, and if so, how?
As an alternative, I could update the row in the Trigger and return a generic error condition, but I think it would be nicer (or at least more intuitive) to have ON DUPLICATE KEY UPDATE work.
REPLACE INTO. Derp! Forgot all about that.
Try below trigger. I am not that good in writing triggers. I have not executed the code below. You can use this as a prototype.
CREATE TRIGGER checkDuplicateEntry
BEFORE INSERT OR UPDATE OF type, ref_id ON employees
DECLARE #duplicateCount INT;
FOR EACH ROW
IF ( new.type <> 'single' )
BEGIN
#duplicateCount = count(*) from yourTableName where type = 'single' AND refid = new.refid
END;
IF ( duplicateCount > 0 )
BEGIN
SIGNAL SQLSTATE VALUE '99999'
SET MESSAGE_TEXT = 'Duplicate Entry';
END