How to ignore consecutive duplicates on insert statement in MySQL - mysql

Question

I want to ignore duplicates (or update key on duplicates), but only on consecutive duplicates.
Say my initial table looks like this (uk and a are column names):
uk a
----
x 1
Iteration 1 inserts 1 so I want this to be ignored in order to avoid duplicates.
Iteration 2 inserts 2 and my table now looks like this:
uk a
----
x 1
x 2
Iteration 3 inserts 1 and because the last row where my unique key = x was different to 1 then I want 1 to be inserted again:
uk a
----
x 1
x 2
x 1
How can this be achieved in MySQL?
My current solution is to query in data from the table first and remove these duplicates, but I would prefer that it was handled by MySQL.

Answer 1

You need a column that will be an auto-increment primary key
CREATE TABLE your_table(
id INT NOT NULL AUTO_INCREMENT,
uk CHAR(30) NOT NULL,
a INT,
PRIMARY KEY (id)
);
INSERT INTO your_table(uk, a) VALUES ('x', 1)
now you can use the following insert command to avoid duplicates as you describe it
INSERT INTO your_table(uk, a)
SELECT 'x', 1 FROM your_table t1
JOIN (
SELECT max(id) maxid, uk
FROM your_table
GROUP BY uk
) t ON t.maxid = t1.id and
t1.a != 1 and
t.uk = 'x'
The second insert can be the following
INSERT INTO your_table(uk, a)
SELECT 'x', 2 FROM your_table t1
JOIN (
SELECT max(id) maxid, uk
FROM your_table
GROUP BY uk
) t ON t.maxid = t1.id and
t1.a != 2 and
t.uk = 'x'
The third insert will be the same as the first one and the result table will be as expected.
See the demo

Answer 2

From your table it is not visible, which values got inserted last. Data in a table is considered unordered. So for your last example we know there are the three records x|1, x|1, x|2 in the table, but not whether x|1 or x|2 was inserted last. You'd need an additional column to indicate this. This could be a datetime or an ascending ID.
If you don't want to change your table, you need a helper table containing the last record instead. Anyway, you'd write a before-insert trigger to look up the last inserted value and throw an error when the new record matches the last inserted record for th uk.
create table mytable(uk char(1), a int);
create table mytable_helper(uk char(1), a int, primary key (uk));
create trigger mytable_ins BEFORE INSERT ON mytable FOR EACH ROW
begin
DECLARE msg VARCHAR(128);
if exists (select * from mytable_helper h where new.uk = h.uk and new.a = h.a) then
set msg = concat('Consecutive duplicate for a = ', new.a, ' and uk = ''', new.uk, '''.');
SIGNAL SQLSTATE '45000' SET MESSAGE_TEXT = msg;
else
replace into mytable_helper (uk, a) values (new.uk, new.a);
end if;
end;
insert into mytable (uk, a) values ('x', 1);
insert into mytable (uk, a) values ('x', 2);
insert into mytable (uk, a) values ('x', 1);
insert into mytable (uk, a) values ('y', 3);
insert into mytable (uk, a) values ('x', 1);
Error(s), warning(s):
Consecutive duplicate for a = 1 and uk = 'x'.
REXTESTER http://rextester.com/XOG65602