I am trying to implement an insert stored procedure. Basically, how it works is the stored procedure will check whether the record exists or not, then proceed to perform insert. A variable #status will be used as indicator. However, I found out a particular problem with this query is that when I execute the query, it will return the result #status = 1 no matter the data existed or not. However, the INSERT function is fine without any problems just the #status. The following is my implementation:
CREATE PROCEDURE save_proc
(
#userid varchar(10)
#name varchar(30)
)
AS
DECLARE #status int
if exists ( SELECT * FROM table1 where userID = #userid AND userName = #name)
SET #status = 0
else
INSERT INTO table1 (userID, userName) VALUES (#userid, #name)
SET #status = 1
SELECT #status
Try to put it in BEGIN END
CREATE PROCEDURE save_proc
(
#userid varchar(10)
#name varchar(30)
)
AS
DECLARE #status int
if exists ( SELECT * FROM table1 where userID = #userid AND userName = #name)
SET #status = 0
else
begin
INSERT INTO table1 (userID, userName) VALUES (#userid, #name)
SET #status = 1
end
If you are not putting the BEGIN END then the scope of your else statement is just the next line which is getting executed ie, INSERT INTO table1 (userID, userName) VALUES (#userid, #name) is only under the scope of else block. And SET #status = 1 is outside the scope of else block. So once the else block executes the next query will be executed which is SET #status = 1
On a side note:
When you are checking for if exists then don't use the * wildcard. Instead you can use 1 i.e,
if exists ( SELECT 1 FROM table1 where userID = #userid AND userName = #name)
else
begin
INSERT INTO table1 (userID, userName) VALUES (#userid, #name)
SET #status = 1
end
Modify code as above. You haven't defined scope of your else. It was upto insert statement only. SET #status = 1 was executed everytime
Related
My stored procedure always returns 0. I tried unique data and duplicated but the insert is done with success but the return value is always the same #new_identity = 0
CREATE PROCEDURE [dbo].[spAddAuthor]
#Author tyAuthor READONLY,
#new_identity INT = NULL OUTPUT
AS
BEGIN
SET NOCOUNT ON;
-- check if the author exists
IF NOT EXISTS (SELECT Id_Author FROM dbo.Authors
WHERE (dbo.Authors.Username = (SELECT Username FROM #Author)
OR dbo.Authors.phone = (SELECT phone FROM #Author)
OR dbo.Authors.email = (SELECT email FROM #Author)))
BEGIN
INSERT INTO dbo.Authors (Username, sexe, email, phone, address)
SELECT [Username], [sexe], [email], [phone], [address]
FROM #Author
-- output the new row
SELECT #new_identity = ##IDENTITY;
END
ELSE
BEGIN
-- get the author Id if already exists
SELECT #new_identity = (SELECT TOP 1 Id_Author
FROM dbo.Authors
WHERE (dbo.Authors.Username = (SELECT Username FROM #Author)
OR dbo.Authors.phone = (SELECT phone FROM #Author)
OR dbo.Authors.email = (SELECT email FROM #Author)))
END
END
I found that in the declaration of the parameters I put null beside the output and that what caused the problem.
#new_identity INT = NULL OUTPUT
but I don't understand why, I thought the 'null' was like the default value, or when you try to make the parameter optional you add null as default value.
can someone explain, please?
I need to validate if a record already exists. The portion, in the stored procedure, that does that is:
SELECT EXISTS (SELECT 1 FROM MyTable
WHERE FirstName = #FirstName
AND LastName = #LastName
AND Address = #Address)
BEGIN
SET #IsNewRecord = 1
END
IF #IsNewRecord = 1
BEGIN
INSERT INTO MyTable
VALUES (#FirtName, #LastName, #Addres, #City, #Phone)
END
That works fine, the issue is when the stored procedure is called by several clients at the same time, it will return the #IsNewRecord = 1.
I already tried surrounding the validation among a TRANSACTION BLOCK, but it still creates a new record.
How can I handle concurrency when using SELECT?
do like this
begin transaction
SELECT EXISTS (SELECT 1 FROM MyTable WHERE FirstName = #FirstName AND LastName = #LastName
AND Address = #Address)
BEGIN
INSERT INTO MyTable
VALUES (#FirtName, #LastName, #Addres, #City, #Phone)
END
commit
I have function below but using it in a select and then inserting the value into temp table is too slow so I try to update to make better performance.
I think it's slow because it selects and then inserts into temp table. If I execute the script without function, it is fast, but I need it to use other place the same table.
CREATE FUNCTION [dbo].[FN_GET_CUSTOMER_INFO]
(#CUSTID VARCHAR(MAX))
RETURNS
#TBL TABLE (CUSTID VARCHAR(200),CUSTNAME NVARCHAR(200))
BEGIN
DECLARE #STATUS VARCHAR(200)
SELECT #STATUS = [STATUS]
FROM TBL_CUSTOMER
WHERE CUSTID = #CUSTID
IF #STATUS = 'INCLUDE'
BEGIN
INSERT INTO #TBL
SELECT *
FROM TBL_CUSTOMER
WHERE CUSTID = #CUSTID
END
ELSE
BEGIN
INSERT INTO #TBL
SELECT *
FROM TBL_CUSTOMER
WHERE CUSTID = #CUSTID
AND GENDER = 'M'
END
END
Can you tell me what is the wrong? I want to have a stored procedure that can do both update and insert action.
CREATE PROCEDURE `save_user`(IN `sp_aliasName` VARCHAR(100),IN `sp_password` VARCHAR(100), IN `sp_sex` VARCHAR(100), IN `sp_age` INT(11), IN `sp_userGroup` VARCHAR(100),OUT `sp_number_of_user` INT)
BEGIN
DECLARE temp INT;
SELECT COUNT(user_id) INTO sp_number_of_user FROM user WHERE aliasName = sp_aliasname;
temp := sp_number_of_user;
IF temp > 0
THEN
UPDATE user SET aliasName=sp_aliasName,password=sp_password,sex=sp_sex,age=sp_age,userGroup=sp_userGroup
WHERE aliasName = sp_aliasname ;
ELSE
INSERT INTO user (aliasName,password,sex,age,userGroup)
values (sp_aliasName,sp_password,sp_sex,sp_age,sp_userGroup);
END IF;
END
I think your procedure should look like this -
CREATE PROCEDURE `save_user`(IN `sp_aliasName` VARCHAR(100),IN `sp_password` VARCHAR(100), IN `sp_sex` VARCHAR(100), IN `sp_age` INT(11), IN `sp_userGroup` VARCHAR(100),OUT `sp_number_of_user` INT)
BEGIN
DECLARE temp INT;
SELECT count(user_id) INTO sp_number_of_user FROM user WHERE aliasName = sp_aliasname;
SET temp := sp_number_of_user;
IF temp > 0 THEN
UPDATE user SET aliasName = sp_aliasName, password = sp_password, sex = sp_sex, age = sp_age, userGroup = sp_userGroup
WHERE aliasName = sp_aliasname;
ELSE
INSERT INTO user (aliasName, password, sex, age, userGroup) VALUES (sp_aliasName, sp_password, sp_sex, sp_age, sp_userGroup);
END IF;
END
I added:
INT type to OUT parameter
WHERE clause to UPDATE statement (check this condition)
Also, have a look at INSERT ... ON DUPLICATE KEY UPDATE command, it may do what you need.
I have a userid table
UserId
JHOSMI
KALVIE
etc...
What I would like to do is create a select statement and pass user id, if the userid already exists then append 1 to the id, This gets complicated if you already have JHOSMI, JHOSMI1, then I want to return JHOSMI2.
Really appreciate help here.
Thanks in advance
edited 21-Jul
this is what i got so far.. but not working the way
select #p AS StaffID,
#old_p := #p,
#Cnt := #Cnt+1 As Lvl,
(SELECT #p :=Concat(#i, #Cnt)
FROM departmenttaff
WHERE upper(trim(UserId)) = upper(trim(StaffID))
AND upper(trim(department)) like upper(trim('SERVICE'))
) AS dummy
FROM (
SELECT
#i := upper(trim('JOHSMI')),
#p := upper(trim('JOHSMI')),
#old_p :='',
#Cnt:=0
) vars,
departmenttaff p
WHERE #p <> #old_p
order by Lvl Desc LIMIT 1;
This will do exactly what you want. You will need a unique constraint on your column.
You might also need to add in error code if success = 0.
This is in MSSQL, you will need to add the relevant commands for MySQL. I do not have MySQL so I cannot test it.
NOTE: You can replace the try catch with some IF EXISTS logic. I just prefer the try catch because its more stable for multiple threads.
begin tran
select * from #tmp
declare #success bit
declare #name varchar(50)
declare #newname varchar(50)
declare #nextid int
declare #attempts int
set #name = 'brad2something'
set #success = 0
set #attempts = 0
while #success = 0 and #attempts < 5 begin
begin try
set #attempts = #attempts + 1 -- failsafe
set #newname = #name
if exists (select * from #tmp where username = #name) begin
select #nextid = isnull(max(convert(int, substring(username, LEN(#name) + 1, 50))), 0) + 1
from #tmp where username like #name + '%' and isnumeric(substring(username, LEN(#name) + 1, 50)) = 1
set #newname = #name + CONVERT(varchar(20), #nextid)
end
insert into #tmp (username) values (#newname)
set #success = 1
end try begin catch end catch
end
--insert into #tmp (username)
--select
select #success
select * from #tmp
rollback
/*
drop table #tmp
create table #tmp (
username varchar(50) not null unique
)
insert into #tmp (username)
select 'brad'
union all select 'brad1'
union all select 'brad2something5'
union all select 'brad2'
union all select 'laney'
union all select 'laney500'
*/
I noticed you want to back fill data. If you want to back fill then this will work. It is extremely inefficient but there is no way around it. There is optimizing code you can put in for when an "error" occurs to prevent all previous counts from happening, but this will work.
begin tran
select * from #tmp
declare #success bit
declare #name varchar(50)
declare #newname varchar(50)
declare #nextid int
declare #attempts int
set #name = 'laney'
set #success = 0
set #attempts = 0
set #nextid = 1
while #success = 0 and #attempts < 5 begin
begin try
if exists (select * from #tmp where username = #name) begin
set #newname = #name + CONVERT(varchar(20), #nextid)
while exists (select * from #tmp where username = #newname) begin
set #nextid = #nextid + 1
set #newname = #name + CONVERT(varchar(20), #nextid)
end
end else
set #newname = #name
set #attempts = #attempts + 1 -- failsafe
insert into #tmp (username) values (#newname)
set #success = 1
end try begin catch end catch
end
--insert into #tmp (username)
--select
select #success
select * from #tmp
rollback
/*
drop table #tmp
create table #tmp (
username varchar(50) not null unique
)
insert into #tmp (username)
select 'brad'
union all select 'brad1'
union all select 'brad2something5'
union all select 'brad2'
union all select 'laney'
union all select 'laney500'
*/
Is it mandatory to have the count in same column? its better to have it in a different integer column. Anyways, if this is the requirement then select userid from table where userid like 'JHOSMI%', then do extract the number using mysql substr function.
For other people who might find this, here's a version in PostgreSQL:
create or replace function uniquify_username(varchar) returns varchar as $$
select $1 || coalesce((max(num) + 1)::varchar, '')
from
(select
substring(name, '^(.*?)[0-9]*$') as prefix,
coalesce(substring(name, '.*([0-9]+)$'), '0')::integer as num
from user1) users
where prefix = $1
$$ LANGUAGE sql;
I think it could be adapted to MySQL (though probably not as a stored procedure) but I don't have a MySQL server handy to do the conversion on.
Put a UNIQUE constraint on the column.
You didn't say what language you are using, so use this pseudo code
counter = 0
finished = false
while finished = false
{
try
{
if counter >= 1 then name = name + counter
counter = counter + 1
insert into table (name)
}
}
This code is extremely finicky. But will get the job done and there is no real other way to do this except for in sql, and you will always have some type of try catch to avoid two processes running at the same time. This way you use the unique key constraint to force the error, and supress it because it is expected.
I in no way condone using try/catch for business logic like this, but you are putting yourself in a situation thats unavoidable. I would say put the ID in a seperate column and make a unique constraint on both fields.
Proper solution:
Columns: Name, ID, Display Name
Unique constraint on: Name, ID
Display Name is a computed column (virtual) is Name + ID
If you do it this way, then all you have to do is INSERT INTO table (name, (select max() from table))