Conditional table updates - mysql

Consider the following table.
myTable
+----+-----------+------------------------------------+
| Id | responseA | responseB |
+----+-----------+------------------------------------+
| 1 | | {"foo":"bar","lvl2":{"key":"val"}} |
+----+-----------+------------------------------------+
where:
Id, INT (11) PRIMARY
responseA, TEXT utf8_unicode_ci
responseB, TEXT utf8_unicode_ci
Let's say that I want to conditionally update the table with some outside data. The conditions are:
• if there's nothing in responseA, populate it with the outside data, otherwise
• if there is something in responseA, leave it as it is, and populate responseB with the outside data
I was pretty much convinced that I could just do this to get what I want:
UPDATE myTable
SET
responseA = IF(TRIM(responseA) = '','foo',TRIM(responseA)),
responseB = IF(TRIM(responseA) != '','foo',TRIM(responseB))
WHERE Id = 1
However, this updates both responseA and responseB to the same value - foo, making the table:
myTable
+----+-----------+-----------+
| Id | responseA | responseB |
+----+-----------+-----------+
| 1 | foo | foo |
+----+-----------+-----------+
I was expecting my table to look like this after the update:
myTable
+----+-----------+------------------------------------+
| Id | responseA | responseB |
+----+-----------+------------------------------------+
| 1 | foo | {"foo":"bar","lvl2":{"key":"val"}} |
+----+-----------+------------------------------------+
What am I misunderstanding, and how can I achieve this conditional update? Do the updates happen sequentially? If so, I guess that would explain why both of the fields are updated.

UPDATE TABLE
SET responseA = CASE WHEN responseA IS NULL
THEN #data
ELSE responseA
END,
responseB = CASE WHEN responseA IS NULL
THEN responseB
ELSE #data
END
;

here your changed query
UPDATE myTable
SET
responseB = IF(TRIM(responseA) != '','foo',TRIM(responseB)),
responseA = IF(TRIM(responseA) = '','foo',TRIM(responseA))
WHERE Id = 1

It seems the value of responseA is changed before the IF() for responseB is evaluated.
One possible solution is to do a simple UPDATE:
UPDATE mytable SET responseA = ? WHERE id = 1
Then adjust the columns in a trigger, where you have access to both the original and the new value of the columns:
CREATE TRIGGER t BEFORE UPDATE ON mytable
FOR EACH ROW BEGIN
IF TRIM(OLD.responseA) != '' THEN
SET NEW.responseB = NEW.responseA;
SET NEW.responseA = OLD.responseA;
END IF;
END
(I have not tested this.)
I am also assuming that your test for '' (empty string) instead of NULL is deliberate, and that you know that NULL is not the same as ''.

The key point in the UPDATE statement is that you should update first the column responseB, so that column responseA retains its original value which can be checked again when you try to update it:
UPDATE myTable
SET responseB = CASE WHEN TRIM(responseA) = '' THEN responseB ELSE 'foo' END,
responseA = CASE WHEN TRIM(responseA) = '' THEN 'foo' ELSE responseA END
WHERE Id = 1;

Related

how to update unique default option in one sql

i have table like this
-----------------------
id | name | is_default|
------------------------
1 | a | 1 |
2 | a | 0 |
3 | a | 0 |
4 | a | 0 |
-----------------------
now i want to change line 2(id =2) is_default to 1,and origin line(id =1) id_default to 0 at the same time,like the choose default option in list in UI.
1.can i do this in one sql statement?
2.if it is possible,how to write the sql statement or how to write in mybatis mapper.xml?
Springboot with mybatis ,sql statement write in mapper.xml
#Data
pulbic class Option{
private Integer id;
private String name;
private Boolean isDefault;
}
how to write the mybatis or mysql statement?
You may use a CASE expression:
UPDATE yourTable
SET is_default = CASE WHEN id = 1 THEN 0 ELSE 1 END
WHERE id IN (1, 2);
Or, if you intended to just toggle the default values for id 1 and 2, then try:
UPDATE yourTable
SET is_default = CASE WHEN is_default = 1 THEN 0 ELSE 1 END
WHERE id IN (1, 2);
Tim's answer is fine. If the values are only 0/1, you can simplify it to:
UPDATE t
SET is_default = 1 - is_default
WHERE id IN (1, 2);

How to select static values on mysql select query?

I am new to mysql, here i am trying to get data from database table.
select id,txnid,amount,status from txn_details;
With above query Getting data successfully but status column getting 0 or 1 or 2, but i want 0 as failed, 1 as success and 2 as not processed.
How to change my query?
You can use a case
select id, txnid, amount,
case when status = 0 then 'failed'
when status = 1 then 'success'
else 'not processed'
end as status
from txn_details;
We can use an expression in the SELECT list. It could be a searched CASE expression e.g.
SELECT CASE t.status
WHEN 0 THEN 'failed'
WHEN 1 THEN 'success'
WHEN 2 THEN 'not processed'
ELSE 'unknown'
END AS status_name
, t.status
, t.amount
, t.txnid
FROM txn_details t
This approach is ANSI-92 standards compliant, and will work in most relational databases.
There are some other MySQL specific alternatives, such as the ELT function ...
SELECT ELT(t.status+1,'failed','success','not processed') AS status_name
, t.status
, t.amount
, t.txnid
FROM txn_details t
https://dev.mysql.com/doc/refman/5.7/en/string-functions.html#function_elt
If you prefer a central point of maintenance (ie you prefer not to recode all your queries when a new status comes along) you could create a status table and either use a join or sub query to get the values, alternatively you could create a function, for example
drop table if exists txn_details,txn_status;
create table txn_details(id int, txnid int, amount int , status int);
insert into txn_details values
(1,1,10,1),(2,1,10,2),(3,1,10,4);
create table txn_status (id int, statusval varchar(20));
insert into txn_status values
(1,'success'),(2,'not processed'), (3,'failed');
drop function if exists f;
delimiter $$
create function f(instatus int)
returns varchar(20)
begin
declare rval varchar(20);
return (select
case when instatus = 1 then 'success'
when instatus = 2 then 'not processed'
when instatus = 3 then 'failed'
else 'Unknown'
end
);
select t.*,coalesce(ts.statusval,'Unknown') status
from txn_details t
left join txn_status ts on ts.id = t.status;
select t.*,coalesce((select statusval from txn_status ts where ts.id = t.status),'Unknown') status
from txn_details t;
Note the use of coalesce in case a status is not found.
Both produce this result
+------+-------+--------+--------+---------------+
| id | txnid | amount | status | status |
+------+-------+--------+--------+---------------+
| 1 | 1 | 10 | 1 | success |
| 2 | 1 | 10 | 2 | not processed |
| 3 | 1 | 10 | 4 | Unknown |
+------+-------+--------+--------+---------------+
3 rows in set (0.00 sec)
Using the function like this
select t.*, f(status) as status
from txn_details t;
also produces the same result.
Of course using a status table or a function means you have to communicate their availability and enforce their use.
I would also consider the using a foreign key constraint in txn_details to cut down on the number of unknown values and put procedures in place to stop people adding new status codes at will without going through change control
The following query would work. It uses CASE ... END to determine and return values for the virtual column status.
SELECT id,txnid,amount,
CASE
WHEN status = 0 THEN 'failed'
WHEN status = 1 THEN 'success'
WHEN status= 2 THEN 'not processed'
END AS status
FROM txn_details;

How to analyze, what happend while updating rows

I have a table like this:
| id | address | name | oid | state | event_id | ctrl |
-------------------------------------------------------------------
| 1 | test_addr_1 | test_1 | 25.345.17 | 1 | 0 | 15 |
I need to get event_id while update data in row.
I want to do something like this:
If new name not equals with old name event_id = event_id + 1
If new oid not equals with old oid event_id = event_id + 2
If new state not equals with old state event_id = event_id + 4
If new ctrl bigger then old ctrl event_id = event_id + 8
# Params to procedure
PROCEDURE Write(IN pAddr VARCHAR(20), IN pName VARCHAR(20), IN pOid VARCHAR(20), IN pState TINYINT, IN pCtrl INT)
#procedure body
SET #ev = 0;
SELECT
CASE
WHEN name != pName THEN SET #ev = #ev + 1
WHEN oid != pOid THEN SET #ev = #ev + 2
WHEN state != pState THEN SET #ev = #ev + 4
WHEN ctrl > pCtrl THEN SET #ev = #ev + 8
END
FROM table1
UPDATE table1 SET ..... , event_id = #ev WHERE address = pAddr
How can I do it? Or will it be better to make it not with the help of SQL?
As already suggested, an audit trigger can be used here to ensure that changes from all sources are caught. However, I suggest two changes:
1) audit tables - use other tables to hold audit data, as these tables tend to grow and it is not recommended to mix operational and auditing data in the same structures (even in the same database)
2) use more friendly change flag - from your example, it seems that you are setting bits in an integer value. While this provides compact data (catch many changes within a single integer), it requires more convoluted operations to see when the name has changed for example. The audit table can simply have BIT(1) columns like nameChanged, oidChanged etc.
CREATE TRIGGER table1Audit BEFORE UPDATE ON <table1>
FOR EACH ROW BEGIN
SET #ev =
(CASE WHEN OLD.name != NEW.name THEN 1 ELSE 0 END) +
(CASE WHEN OLD.oid != NEW.oid THEN 2 ELSE 0 END) +
(CASE WHEN OLD.state != NEW.state THEN 4 ELSE 0 END) +
(CASE WHEN OLD.ctrl != NEW.ctrl THEN 8 ELSE 0 END)
-- INSERT INTO someaudittable
--'table1', #ev
END;
Solved! Thanks to #Alexei.
I wanted to know, why the row was added to the history table.
The result is:
CREATE TRIGGER SetReason BEFORE UPDATE ON <table1>
FOR EACH ROW BEGIN
SET NEW.event_id =
(CASE WHEN OLD.name != NEW.name THEN 1 ELSE 0 END) +
(CASE WHEN OLD.oid != NEW.oid THEN 2 ELSE 0 END) +
(CASE WHEN OLD.state != NEW.state THEN 4 ELSE 0 END) +
(CASE WHEN OLD.ctrl != NEW.ctrl THEN 8 ELSE 0 END)
END;

How to check record isn't containing specific string?

I have a table like this:
// mytable
+----+-------------------------------------------+
| id | col |
+----+-------------------------------------------+
| 1 | Peter|423421 , Alex Jon|61333 |
| 2 | Barmar|624321 |
| 3 | Jack|624321 , Ali|312331 , Leonard|624321 |
+----+-------------------------------------------+
I need to check second row if isn't containing 824326, then add this value , Sara|824326 in the end of that. Something like this:
| 2 | Barmar|624321 , Sara|824326 |
Now I want to know, how can I check that field isn't containing thins number 824326 ?
Here is my try, I just need to the condition (to check existing)
UPDATE mytable
SET col = CASE WHEN col = {:val doesn't exist} // <= How to write this condition?
THEN CONCAT(col,':newval')
ELSE SIGNAL SQLSTATE '45000' SET MESSAGE_TEXT = "It is duplicate";
END,
WHERE id = 2;
Note1: :newval is this , Sara|824326 in the above query.
Note2: :val is this 824326 in the above query.
You have to use INSTR. Try this:
UPDATE mytable
SET col = CASE WHEN col INSTR(col, ':val') = 0
THEN CONCAT(col,':newval')
ELSE SIGNAL SQLSTATE '45000' SET MESSAGE_TEXT = "It is duplicate";
END,
WHERE id = 2;

How to update in SQL based on different where clause

I have a table which has a structure as:
ID DBInstance DBName Tag
1 | INS1 | master | NULL
2 | INS1 | tempdb | NULL
4 | INS2 | master | NULL
5 | INS2 | tempdb | NULL
I want to update the tag in this table as based on the condition as:
1) Update tag as "a" only for DBInstance as "INS1" and DBName as "master"
2) Update tag as "b" only for DBInstance as "INS2" and DBName as "tempdb".
And I want to update both of these two only in a single statement, not in two different update queries. How can I do so?
A query somewhat like this:
UPDATE tbl_test
SET tag = 'a' where DBInstance in ('INS1') and DBName IN ('master'),
tag = 'b' where DBInstance in ('INS2') and DBName IN ('tempdb')
But obviously, this query is wrong, so how can I do so?
You can use the case when then like this:
UPDATE tbl_test
SET tag = case
when DBInstance = 'INS1' and DBName = 'master'
then 'a'
when DBInstance = 'INS2' and DBName = 'tempdb'
then 'b'
else NULL -- or may be tag, or default value which you want.
end