SQL Server Insert if not exists

Question

I want to insert data into my table, but insert only data that doesn't already exist in my database.

Here is my code:

ALTER PROCEDURE [dbo].[EmailsRecebidosInsert]
  (@_DE nvarchar(50),
   @_ASSUNTO nvarchar(50),
   @_DATA nvarchar(30) )
AS
BEGIN
   INSERT INTO EmailsRecebidos (De, Assunto, Data)
   VALUES (@_DE, @_ASSUNTO, @_DATA)
   WHERE NOT EXISTS ( SELECT * FROM EmailsRecebidos 
                   WHERE De = @_DE
                   AND Assunto = @_ASSUNTO
                   AND Data = @_DATA);
END

And the error is:

Msg 156, Level 15, State 1, Procedure EmailsRecebidosInsert, Line 11
Incorrect syntax near the keyword 'WHERE'.

Answer 1

instead of below Code

BEGIN
   INSERT INTO EmailsRecebidos (De, Assunto, Data)
   VALUES (@_DE, @_ASSUNTO, @_DATA)
   WHERE NOT EXISTS ( SELECT * FROM EmailsRecebidos 
                   WHERE De = @_DE
                   AND Assunto = @_ASSUNTO
                   AND Data = @_DATA);
END

replace with

BEGIN
   IF NOT EXISTS (SELECT * FROM EmailsRecebidos 
                   WHERE De = @_DE
                   AND Assunto = @_ASSUNTO
                   AND Data = @_DATA)
   BEGIN
       INSERT INTO EmailsRecebidos (De, Assunto, Data)
       VALUES (@_DE, @_ASSUNTO, @_DATA)
   END
END

Updated : (thanks to @Marc Durdin for pointing)

Note that under high load, this will still sometimes fail, because a second connection can pass the IF NOT EXISTS test before the first connection executes the INSERT, i.e. a race condition. See stackoverflow.com/a/3791506/1836776 for a good answer on why even wrapping in a transaction doesn't solve this.

Answer 2

For those looking for the fastest way, I recently came across these benchmarks where apparently using "INSERT SELECT... EXCEPT SELECT..." turned out to be the fastest for 50 million records or more.

Here's some sample code from the article (the 3rd block of code was the fastest):

INSERT INTO #table1 (Id, guidd, TimeAdded, ExtraData)
SELECT Id, guidd, TimeAdded, ExtraData
FROM #table2
WHERE NOT EXISTS (Select Id, guidd From #table1 WHERE #table1.id = #table2.id)
-----------------------------------
MERGE #table1 as [Target]
USING  (select Id, guidd, TimeAdded, ExtraData from #table2) as [Source]
(id, guidd, TimeAdded, ExtraData)
    on [Target].id =[Source].id
WHEN NOT MATCHED THEN
    INSERT (id, guidd, TimeAdded, ExtraData)
    VALUES ([Source].id, [Source].guidd, [Source].TimeAdded, [Source].ExtraData);
------------------------------
INSERT INTO #table1 (id, guidd, TimeAdded, ExtraData)
SELECT id, guidd, TimeAdded, ExtraData from #table2
EXCEPT
SELECT id, guidd, TimeAdded, ExtraData from #table1
------------------------------
INSERT INTO #table1 (id, guidd, TimeAdded, ExtraData)
SELECT #table2.id, #table2.guidd, #table2.TimeAdded, #table2.ExtraData
FROM #table2
LEFT JOIN #table1 on #table1.id = #table2.id
WHERE #table1.id is null

Answer 3

I would use a merge:

create PROCEDURE [dbo].[EmailsRecebidosInsert]
  (@_DE nvarchar(50),
   @_ASSUNTO nvarchar(50),
   @_DATA nvarchar(30) )
AS
BEGIN
   with data as (select @_DE as de, @_ASSUNTO as assunto, @_DATA as data)
   merge EmailsRecebidos t
   using data s
      on s.de = t.de
     and s.assunte = t.assunto
     and s.data = t.data
    when not matched by target
    then insert (de, assunto, data) values (s.de, s.assunto, s.data);
END

Answer 4

Try below code

ALTER PROCEDURE [dbo].[EmailsRecebidosInsert]
  (@_DE nvarchar(50),
   @_ASSUNTO nvarchar(50),
   @_DATA nvarchar(30) )
AS
BEGIN
   INSERT INTO EmailsRecebidos (De, Assunto, Data)
   select @_DE, @_ASSUNTO, @_DATA
   EXCEPT
   SELECT De, Assunto, Data from EmailsRecebidos
END

Answer 5

I did the same thing with SQL Server 2012 and it worked

Insert into #table1 With (ROWLOCK) (Id, studentId, name)
SELECT '18769', '2', 'Alex'
WHERE not exists (select * from #table1 where Id = '18769' and studentId = '2')

Answer 6

Different SQL, same principle. Only insert if the clause in where not exists fails

INSERT INTO FX_USDJPY
            (PriceDate, 
            PriceOpen, 
            PriceLow, 
            PriceHigh, 
            PriceClose, 
            TradingVolume, 
            TimeFrame)
    SELECT '2014-12-26 22:00',
           120.369000000000,
           118.864000000000,
           120.742000000000,
           120.494000000000,
           86513,
           'W'
    WHERE NOT EXISTS
        (SELECT 1
         FROM FX_USDJPY
         WHERE PriceDate = '2014-12-26 22:00'
           AND TimeFrame = 'W')

Answer 7

The INSERT command doesn't have a WHERE clause - you'll have to write it like this:

ALTER PROCEDURE [dbo].[EmailsRecebidosInsert]
  (@_DE nvarchar(50),
   @_ASSUNTO nvarchar(50),
   @_DATA nvarchar(30) )
AS
BEGIN
   IF NOT EXISTS (SELECT * FROM EmailsRecebidos 
                   WHERE De = @_DE
                   AND Assunto = @_ASSUNTO
                   AND Data = @_DATA)
   BEGIN
       INSERT INTO EmailsRecebidos (De, Assunto, Data)
       VALUES (@_DE, @_ASSUNTO, @_DATA)
   END
END

Answer 8

Depending on your version (2012?) of SQL Server aside from the IF EXISTS you can also use MERGE like so:

ALTER PROCEDURE [dbo].[EmailsRecebidosInsert]
    ( @_DE nvarchar(50)
    , @_ASSUNTO nvarchar(50)
    , @_DATA nvarchar(30))
AS BEGIN
    MERGE [dbo].[EmailsRecebidos] [Target]
    USING (VALUES (@_DE, @_ASSUNTO, @_DATA)) [Source]([De], [Assunto], [Data])
         ON [Target].[De] = [Source].[De] AND [Target].[Assunto] = [Source].[Assunto] AND [Target].[Data] = [Source].[Data]
     WHEN NOT MATCHED THEN
        INSERT ([De], [Assunto], [Data])
        VALUES ([Source].[De], [Source].[Assunto], [Source].[Data]);
END

Answer 9

If your clustered index consist from only those fields than the simple, fast and reliable option is to use IGNORE_DUP_KEY

If you create the Clustered index with IGNORE_DUP_KEY ON

Than you can just use:

INSERT INTO EmailsRecebidos (De, Assunto, Data) VALUES (@_DE, @_ASSUNTO, @_DATA)

This should be safe in all cases!

Answer 10

As explained in below code: Execute below queries and verify yourself.

CREATE TABLE `table_name` (
  `id` int(11) NOT NULL auto_increment,
  `name` varchar(255) NOT NULL,
  `address` varchar(255) NOT NULL,
  `tele` varchar(255) NOT NULL,
  PRIMARY KEY  (`id`)
) ENGINE=InnoDB;

Insert a record:

INSERT INTO table_name (name, address, tele)
SELECT * FROM (SELECT 'Nazir', 'Kolkata', '033') AS tmp
WHERE NOT EXISTS (
    SELECT name FROM table_name WHERE name = 'Nazir'
) LIMIT 1;
Query OK, 1 row affected (0.00 sec)
Records: 1 Duplicates: 0 Warnings: 0

SELECT * FROM `table_name`;

+----+--------+-----------+------+
| id | name   | address   | tele |
+----+--------+-----------+------+
|  1 | Nazir  | Kolkata   | 033  |
+----+--------+-----------+------+

Now, try to insert the same record again:

INSERT INTO table_name (name, address, tele)
SELECT * FROM (SELECT 'Nazir', 'Kolkata', '033') AS tmp
WHERE NOT EXISTS (
    SELECT name FROM table_name WHERE name = 'Nazir'
) LIMIT 1;

Query OK, 0 rows affected (0.00 sec)
Records: 0  Duplicates: 0  Warnings: 0

+----+--------+-----------+------+
| id | name   | address   | tele |
+----+--------+-----------+------+
|  1 | Nazir  | Kolkata   | 033  |
+----+--------+-----------+------+

Insert a different record:

INSERT INTO table_name (name, address, tele)
SELECT * FROM (SELECT 'Santosh', 'Kestopur', '044') AS tmp
WHERE NOT EXISTS (
    SELECT name FROM table_name WHERE name = 'Santosh'
) LIMIT 1;

Query OK, 1 row affected (0.00 sec)
Records: 1 Duplicates: 0 Warnings: 0

SELECT * FROM `table_name`;

+----+--------+-----------+------+
| id | name   | address   | tele |
+----+--------+-----------+------+
|  1 | Nazir  | Kolkata   | 033  |
|  2 | Santosh| Kestopur  | 044  |
+----+--------+-----------+------+

Answer 11

You could use the GO command. That will restart the execution of SQL statements after an error. In my case I have a few 1000 INSERT statements, where a handful of those records already exist in the database, I just don't know which ones. I found that after processing a few 100, execution just stops with an error message that it can't INSERT as the record already exists. Quite annoying, but putting a GO solved this. It may not be the fastest solution, but speed was not my problem.

GO
INSERT INTO mytable (C1,C2,C3) VALUES(1,2,3)
GO
INSERT INTO mytable (C1,C2,C3) VALUES(4,5,6)
 etc ...