I am trying to perform a merge into a table (let's call it table1) from a table2. In the USING condition I need a third table (table3). This third table contains some IDs that I need in table1. A simplified version of my merge looks like:
MERGE INTO table1 a
USING (
SELECT ID, address
FROM table3 b
Where address IN
(
SELECT address
FROM table3
WHERE address IS NOT NULL
AND ID> 0
GROUP BY address
HAVING COUNT(*) = 1
)
) c
ON (a.address = c.address)
WHEN MATCHED THEN
UPDATE SET a.ID = c.ID
WHERE a.ID = 0
I know that the error I get is usually caused by the query in the USING clause, but theoretically this problem should be eliminated by the count(*)=1 condition.
I have duplicates in table2, but they should all get an ID from table3 or ID 0 if the address is duplicated in table3.
IDs are unique for an address, so they should be distinct.
P.S. This merge is performed automatically by a script that , so I can modify the query to add more conditions/restrictions, but I cannot change the structure [meaning I have to use these 3 tables as they are].
I hope this makes sense.
Any ideas why this still does not work for me?
Try this:
MERGE INTO table1 a
USING (
SELECT max(ID), address
FROM table3 b
WHERE address IS NOT NULL AND ID > 0
GROUP BY address
HAVING COUNT(*) = 1
) c
ON (a.address = c.address)
WHEN MATCHED THEN
UPDATE SET a.ID = c.ID
WHERE a.ID = 0;
you have where condition in inner query but not in outer query. If you want your original query please try:
MERGE INTO table1 a
USING (
SELECT ID, address
FROM table3 b
AND address IN
(
SELECT address
FROM table3
WHERE address IS NOT NULL
AND ID> 0
GROUP BY address
HAVING COUNT(*) = 1
)
WHERE address IS NOT NULL
AND ID> 0
) c
ON (a.address = c.address)
WHEN MATCHED THEN
UPDATE SET a.ID = c.ID
WHERE a.ID = 0
The issue is more than likely due to the duplicate rows from table2. Here's a simple test case demonstrating the issue:
Setup:
CREATE TABLE t1 (ID INTEGER PRIMARY KEY,
val VARCHAR2(1));
CREATE TABLE t2 (ID INTEGER,
val VARCHAR2(1));
INSERT INTO t1 (ID, val) VALUES (1, 'A');
INSERT INTO t2 (ID, val) VALUES (1, 'B');
INSERT INTO t2 (ID, val) VALUES (1, 'B');
COMMIT;
Merge that will error:
MERGE INTO t1 USING t2
ON (t1.id = t2.id)
WHEN MATCHED THEN
UPDATE SET t1.val = t2.val;
ORA-30926: unable to get a stable set of rows in the source tables
Merge that will succeed:
MERGE INTO t1 USING (SELECT DISTINCT id, val FROM t2) t2
ON (t1.id = t2.id)
WHEN MATCHED THEN
UPDATE SET t1.val = t2.val;
N.B. The second merge will still fail if you have different values returned for val for the same id; that means you will have more than one row returned for a given id, and Oracle won't know which one to use to update the target table with.
In order to make sure your merge statement will work, you will need to ensure that you will return at most 1 row per address in the source subquery.
I need to copy a lot of data from one table to another. If the data already exists, I need to update it, otherwise I need to insert it. The data to be copied is selecting using a WHERE condition. The data has a primary key (a string of up to 12 characters).
If I was just inserting the data, I would do
INSERT INTO T2 SELECT COL1, COL2 FROM T1 WHERE T1.ID ='I'
but I cannot figure out how to do the INSERT / UPDATE. I keep seeing references to upserts and MERGE, but MERGE appears to have issues,and I cannot figure ut how to do the upsert for multiple records.
What is the best solution for this?
If you want to avoid merges (though you should not be afraid of it) you can do something like
update t2
set col1 = t1.col1
,col2 = t1.col2
from t2
join t1
on t2.[joinkey] = t1.[joinkey]
where [where clause]
And after for the ones that you do not have
insert into t2(col1,col2)
select col1,col2 from t1
where not exists (select * from t2 where t1.[joinkey] = t2.[joinkey])
in such way you first update the ones that match and then insert the ones that do not. Also if you want it in one go you can wrap it in a transaction.
It is commonly known as UPSERT operation. Yes you are correct in saying merge has some issues with it so stay away from it.
A simple approach assuming there is a Primary Key column in Both tables called PK_Col would be something like this...
BEGIN TRANSACTION;
-- Update already existing records
UPDATE T2
SET T2.Col1 = T1.Col1
,T2.Col2 = T1.Col2
FROM T2 INNER JOIN T1 ON T2.PK_COl = T1.PK_Col
-- Insert missing records
INSERT INTO T2 (COL1, COL2 )
SELECT COL1, COL2
FROM T1
WHERE T1.ID ='I'
AND NOT EXISTS (SELECT 1
FROM T2
WHERE T2.PK_COl = T1.PK_Col )
COMMIT TRANSACTION;
Wrap the whole UPSERT operation in one transaction.
You can use IF EXISTS something like:
if exists (select * from table with (updlock,serializable) where key = #key)
begin
update table set ...
where key = #key
end
else
begin
insert table (key, ...)
values (#key, ...)
end
Another solution is to check ##ROWCOUNT
UPDATE MyTable SET FieldA=#FieldA WHERE Key=#Key
IF ##ROWCOUNT = 0
INSERT INTO MyTable (FieldA) VALUES (#FieldA)
I have the following two tables:
Table1
----------
ID Name
1 A
2 B
3 C
Table2
----------
ID Name
1 Z
I need to insert data from Table1 to Table2. I can use the following syntax:
INSERT INTO Table2(Id, Name) SELECT Id, Name FROM Table1
However, in my case, duplicate IDs might exist in Table2 (in my case, it's just "1") and I don't want to copy that again as that would throw an error.
I can write something like this:
IF NOT EXISTS(SELECT 1 FROM Table2 WHERE Id=1)
INSERT INTO Table2 (Id, name) SELECT Id, name FROM Table1
ELSE
INSERT INTO Table2 (Id, name) SELECT Id, name FROM Table1 WHERE Table1.Id<>1
Is there a better way to do this without using IF - ELSE? I want to avoid two INSERT INTO-SELECT statements based on some condition.
Using NOT EXISTS:
INSERT INTO TABLE_2
(id, name)
SELECT t1.id,
t1.name
FROM TABLE_1 t1
WHERE NOT EXISTS(SELECT id
FROM TABLE_2 t2
WHERE t2.id = t1.id)
Using NOT IN:
INSERT INTO TABLE_2
(id, name)
SELECT t1.id,
t1.name
FROM TABLE_1 t1
WHERE t1.id NOT IN (SELECT id
FROM TABLE_2)
Using LEFT JOIN/IS NULL:
INSERT INTO TABLE_2
(id, name)
SELECT t1.id,
t1.name
FROM TABLE_1 t1
LEFT JOIN TABLE_2 t2 ON t2.id = t1.id
WHERE t2.id IS NULL
Of the three options, the LEFT JOIN/IS NULL is less efficient. See this link for more details.
In MySQL you can do this:
INSERT IGNORE INTO Table2(Id, Name) SELECT Id, Name FROM Table1
Does SQL Server have anything similar?
I just had a similar problem, the DISTINCT keyword works magic:
INSERT INTO Table2(Id, Name) SELECT DISTINCT Id, Name FROM Table1
I was facing the same problem recently...
Heres what worked for me in MS SQL server 2017...
The primary key should be set on ID in table 2...
The columns and column properties should be the same of course between both tables. This will work the first time you run the below script. The duplicate ID in table 1, will not insert...
If you run it the second time, you will get a
Violation of PRIMARY KEY constraint error
This is the code:
Insert into Table_2
Select distinct *
from Table_1
where table_1.ID >1
Using ignore Duplicates on the unique index as suggested by IanC here was my solution for a similar issue, creating the index with the Option WITH IGNORE_DUP_KEY
In backward compatible syntax
, WITH IGNORE_DUP_KEY is equivalent to WITH IGNORE_DUP_KEY = ON.
Ref.: index_option
From SQL Server you can set a Unique key index on the table for (Columns that needs to be unique)
A little off topic, but if you want to migrate the data to a new table, and the possible duplicates are in the original table, and the column possibly duplicated is not an id, a GROUP BY will do:
INSERT INTO TABLE_2
(name)
SELECT t1.name
FROM TABLE_1 t1
GROUP BY t1.name
In my case, I had duplicate IDs in the source table, so none of the proposals worked. I don't care about performance, it's just done once.
To solve this I took the records one by one with a cursor to ignore the duplicates.
So here's the code example:
DECLARE #c1 AS VARCHAR(12);
DECLARE #c2 AS VARCHAR(250);
DECLARE #c3 AS VARCHAR(250);
DECLARE MY_cursor CURSOR STATIC FOR
Select
c1,
c2,
c3
from T2
where ....;
OPEN MY_cursor
FETCH NEXT FROM MY_cursor INTO #c1, #c2, #c3
WHILE ##FETCH_STATUS = 0
BEGIN
if (select count(1)
from T1
where a1 = #c1
and a2 = #c2
) = 0
INSERT INTO T1
values (#c1, #c2, #c3)
FETCH NEXT FROM MY_cursor INTO #c1, #c2, #c3
END
CLOSE MY_cursor
DEALLOCATE MY_cursor
I used a MERGE query to fill a table without duplications.
The problem I had was a double key in the tables ( Code , Value ) ,
and the exists query was very slow
The MERGE executed very fast ( more then X100 )
examples for MERGE query
For one table it works perfectly when creating one unique index from multiple field. Then simple "INSERT IGNORE" will ignore duplicates if ALL of 7 fields (in this case) will have SAME values.
Select fields in PMA Structure View and click Unique, new combined index will be created.
A simple DELETE before the INSERT would suffice:
DELETE FROM Table2 WHERE Id = (SELECT Id FROM Table1)
INSERT INTO Table2 (Id, name) SELECT Id, name FROM Table1
Switching Table1 for Table2 depending on which table's Id and name pairing you want to preserve.
In SQL Server, it is possible to insert rows into a table with an INSERT.. SELECT statement:
INSERT INTO Table (col1, col2, col3)
SELECT col1, col2, col3
FROM other_table
WHERE sql = 'cool'
Is it also possible to update a table with SELECT? I have a temporary table containing the values and would like to update another table using those values. Perhaps something like this:
UPDATE Table SET col1, col2
SELECT col1, col2
FROM other_table
WHERE sql = 'cool'
WHERE Table.id = other_table.id
UPDATE
Table_A
SET
Table_A.col1 = Table_B.col1,
Table_A.col2 = Table_B.col2
FROM
Some_Table AS Table_A
INNER JOIN Other_Table AS Table_B
ON Table_A.id = Table_B.id
WHERE
Table_A.col3 = 'cool'
In SQL Server 2008 (or newer), use MERGE
MERGE INTO YourTable T
USING other_table S
ON T.id = S.id
AND S.tsql = 'cool'
WHEN MATCHED THEN
UPDATE
SET col1 = S.col1,
col2 = S.col2;
Alternatively:
MERGE INTO YourTable T
USING (
SELECT id, col1, col2
FROM other_table
WHERE tsql = 'cool'
) S
ON T.id = S.id
WHEN MATCHED THEN
UPDATE
SET col1 = S.col1,
col2 = S.col2;
UPDATE YourTable
SET Col1 = OtherTable.Col1,
Col2 = OtherTable.Col2
FROM (
SELECT ID, Col1, Col2
FROM other_table) AS OtherTable
WHERE
OtherTable.ID = YourTable.ID
I'd modify Robin's excellent answer to the following:
UPDATE Table
SET Table.col1 = other_table.col1,
Table.col2 = other_table.col2
FROM
Table
INNER JOIN other_table ON Table.id = other_table.id
WHERE
Table.col1 != other_table.col1
OR Table.col2 != other_table.col2
OR (
other_table.col1 IS NOT NULL
AND Table.col1 IS NULL
)
OR (
other_table.col2 IS NOT NULL
AND Table.col2 IS NULL
)
Without a WHERE clause, you'll affect even rows that don't need to be affected, which could (possibly) cause index recalculation or fire triggers that really shouldn't have been fired.
One way
UPDATE t
SET t.col1 = o.col1,
t.col2 = o.col2
FROM
other_table o
JOIN
t ON t.id = o.id
WHERE
o.sql = 'cool'
Another possibility not mentioned yet is to just chuck the SELECT statement itself into a CTE and then update the CTE.
WITH CTE
AS (SELECT T1.Col1,
T2.Col1 AS _Col1,
T1.Col2,
T2.Col2 AS _Col2
FROM T1
JOIN T2
ON T1.id = T2.id
/*Where clause added to exclude rows that are the same in both tables
Handles NULL values correctly*/
WHERE EXISTS(SELECT T1.Col1,
T1.Col2
EXCEPT
SELECT T2.Col1,
T2.Col2))
UPDATE CTE
SET Col1 = _Col1,
Col2 = _Col2;
This has the benefit that it is easy to run the SELECT statement on its own first to sanity check the results, but it does requires you to alias the columns as above if they are named the same in source and target tables.
This also has the same limitation as the proprietary UPDATE ... FROM syntax shown in four of the other answers. If the source table is on the many side of a one-to-many join then it is undeterministic which of the possible matching joined records will be used in the Update (an issue that MERGE avoids by raising an error if there is an attempt to update the same row more than once).
For the record (and others searching like I was), you can do it in MySQL like this:
UPDATE first_table, second_table
SET first_table.color = second_table.color
WHERE first_table.id = second_table.foreign_id
Using alias:
UPDATE t
SET t.col1 = o.col1
FROM table1 AS t
INNER JOIN
table2 AS o
ON t.id = o.id
The simple way to do it is:
UPDATE
table_to_update,
table_info
SET
table_to_update.col1 = table_info.col1,
table_to_update.col2 = table_info.col2
WHERE
table_to_update.ID = table_info.ID
This may be a niche reason to perform an update (for example, mainly used in a procedure), or may be obvious to others, but it should also be stated that you can perform an update-select statement without using join (in case the tables you're updating between have no common field).
update
Table
set
Table.example = a.value
from
TableExample a
where
Table.field = *key value* -- finds the row in Table
AND a.field = *key value* -- finds the row in TableExample a
Here is another useful syntax:
UPDATE suppliers
SET supplier_name = (SELECT customers.name
FROM customers
WHERE customers.customer_id = suppliers.supplier_id)
WHERE EXISTS (SELECT customers.name
FROM customers
WHERE customers.customer_id = suppliers.supplier_id);
It checks if it is null or not by using "WHERE EXIST".
I add this only so you can see a quick way to write it so that you can check what will be updated before doing the update.
UPDATE Table
SET Table.col1 = other_table.col1,
Table.col2 = other_table.col2
--select Table.col1, other_table.col,Table.col2,other_table.col2, *
FROM Table
INNER JOIN other_table
ON Table.id = other_table.id
If you use MySQL instead of SQL Server, the syntax is:
UPDATE Table1
INNER JOIN Table2
ON Table1.id = Table2.id
SET Table1.col1 = Table2.col1,
Table1.col2 = Table2.col2
UPDATE from SELECT with INNER JOIN in SQL Database
Since there are too many replies of this post, which are most heavily up-voted, I thought I would provide my suggestion here too. Although the question is very interesting, I have seen in many forum sites and made a solution using INNER JOIN with screenshots.
At first, I have created a table named with schoolold and inserted few records with respect to their column names and execute it.
Then I executed SELECT command to view inserted records.
Then I created a new table named with schoolnew and similarly executed above actions on it.
Then, to view inserted records in it, I execute SELECT command.
Now, Here I want to make some changes in third and fourth row, to complete this action, I execute UPDATE command with INNER JOIN.
To view the changes I execute the SELECT command.
You can see how Third and Fourth records of table schoolold easily replaced with table schoolnew by using INNER JOIN with UPDATE statement.
And if you wanted to join the table with itself (which won't happen too often):
update t1 -- just reference table alias here
set t1.somevalue = t2.somevalue
from table1 t1 -- these rows will be the targets
inner join table1 t2 -- these rows will be used as source
on .................. -- the join clause is whatever suits you
Updating through CTE is more readable than the other answers here:
;WITH cte
AS (SELECT col1,col2,id
FROM other_table
WHERE sql = 'cool')
UPDATE A
SET A.col1 = B.col1,
A.col2 = B.col2
FROM table A
INNER JOIN cte B
ON A.id = B.id
The following example uses a derived table, a SELECT statement after the FROM clause, to return the old and new values for further updates:
UPDATE x
SET x.col1 = x.newCol1,
x.col2 = x.newCol2
FROM (SELECT t.col1,
t2.col1 AS newCol1,
t.col2,
t2.col2 AS newCol2
FROM [table] t
JOIN other_table t2
ON t.ID = t2.ID) x
If you are using SQL Server you can update one table from another without specifying a join and simply link the two from the where clause. This makes a much simpler SQL query:
UPDATE Table1
SET Table1.col1 = Table2.col1,
Table1.col2 = Table2.col2
FROM
Table2
WHERE
Table1.id = Table2.id
Consolidating all the different approaches here.
Select update
Update with a common table expression
Merge
Sample table structure is below and will update from Product_BAK to Product table.
Table Product
CREATE TABLE [dbo].[Product](
[Id] [int] IDENTITY(1, 1) NOT NULL,
[Name] [nvarchar](100) NOT NULL,
[Description] [nvarchar](100) NULL
) ON [PRIMARY]
Table Product_BAK
CREATE TABLE [dbo].[Product_BAK](
[Id] [int] IDENTITY(1, 1) NOT NULL,
[Name] [nvarchar](100) NOT NULL,
[Description] [nvarchar](100) NULL
) ON [PRIMARY]
1. Select update
update P1
set Name = P2.Name
from Product P1
inner join Product_Bak P2 on p1.id = P2.id
where p1.id = 2
2. Update with a common table expression
; With CTE as
(
select id, name from Product_Bak where id = 2
)
update P
set Name = P2.name
from product P inner join CTE P2 on P.id = P2.id
where P2.id = 2
3. Merge
Merge into product P1
using Product_Bak P2 on P1.id = P2.id
when matched then
update set p1.[description] = p2.[description], p1.name = P2.Name;
In this Merge statement, we can do insert if not finding a matching record in the target, but exist in the source and please find syntax:
Merge into product P1
using Product_Bak P2 on P1.id = P2.id;
when matched then
update set p1.[description] = p2.[description], p1.name = P2.Name;
WHEN NOT MATCHED THEN
insert (name, description)
values(p2.name, P2.description);
The other way is to use a derived table:
UPDATE t
SET t.col1 = a.col1
,t.col2 = a.col2
FROM (
SELECT id, col1, col2 FROM #tbl2) a
INNER JOIN #tbl1 t ON t.id = a.id
Sample data
DECLARE #tbl1 TABLE (id INT, col1 VARCHAR(10), col2 VARCHAR(10))
DECLARE #tbl2 TABLE (id INT, col1 VARCHAR(10), col2 VARCHAR(10))
INSERT #tbl1 SELECT 1, 'a', 'b' UNION SELECT 2, 'b', 'c'
INSERT #tbl2 SELECT 1, '1', '2' UNION SELECT 2, '3', '4'
UPDATE t
SET t.col1 = a.col1
,t.col2 = a.col2
FROM (
SELECT id, col1, col2 FROM #tbl2) a
INNER JOIN #tbl1 t ON t.id = a.id
SELECT * FROM #tbl1
SELECT * FROM #tbl2
UPDATE TQ
SET TQ.IsProcessed = 1, TQ.TextName = 'bla bla bla'
FROM TableQueue TQ
INNER JOIN TableComment TC ON TC.ID = TQ.TCID
WHERE TQ.IsProcessed = 0
To make sure you are updating what you want, select first
SELECT TQ.IsProcessed, 1 AS NewValue1, TQ.TextName, 'bla bla bla' AS NewValue2
FROM TableQueue TQ
INNER JOIN TableComment TC ON TC.ID = TQ.TCID
WHERE TQ.IsProcessed = 0
There is even a shorter method and it might be surprising for you:
Sample data set:
CREATE TABLE #SOURCE ([ID] INT, [Desc] VARCHAR(10));
CREATE TABLE #DEST ([ID] INT, [Desc] VARCHAR(10));
INSERT INTO #SOURCE VALUES(1,'Desc_1'), (2, 'Desc_2'), (3, 'Desc_3');
INSERT INTO #DEST VALUES(1,'Desc_4'), (2, 'Desc_5'), (3, 'Desc_6');
Code:
UPDATE #DEST
SET #DEST.[Desc] = #SOURCE.[Desc]
FROM #SOURCE
WHERE #DEST.[ID] = #SOURCE.[ID];
Use:
drop table uno
drop table dos
create table uno
(
uid int,
col1 char(1),
col2 char(2)
)
create table dos
(
did int,
col1 char(1),
col2 char(2),
[sql] char(4)
)
insert into uno(uid) values (1)
insert into uno(uid) values (2)
insert into dos values (1,'a','b',null)
insert into dos values (2,'c','d','cool')
select * from uno
select * from dos
EITHER:
update uno set col1 = (select col1 from dos where uid = did and [sql]='cool'),
col2 = (select col2 from dos where uid = did and [sql]='cool')
OR:
update uno set col1=d.col1,col2=d.col2 from uno
inner join dos d on uid=did where [sql]='cool'
select * from uno
select * from dos
If the ID column name is the same in both tables then just put the table name before the table to be updated and use an alias for the selected table, i.e.:
update uno set col1 = (select col1 from dos d where uno.[id] = d.[id] and [sql]='cool'),
col2 = (select col2 from dos d where uno.[id] = d.[id] and [sql]='cool')
In the accepted answer, after the:
SET
Table_A.col1 = Table_B.col1,
Table_A.col2 = Table_B.col2
I would add:
OUTPUT deleted.*, inserted.*
What I usually do is putting everything in a roll backed transaction and using the "OUTPUT": in this way I see everything that is about to happen. When I am happy with what I see, I change the ROLLBACK into COMMIT.
I usually need to document what I did, so I use the "results to Text" option when I run the roll-backed query and I save both the script and the result of the OUTPUT. (Of course this is not practical if I changed too many rows)
UPDATE table AS a
INNER JOIN table2 AS b
ON a.col1 = b.col1
INNER JOIN ... AS ...
ON ... = ...
SET ...
WHERE ...
The below solution works for a MySQL database:
UPDATE table1 a , table2 b
SET a.columname = 'some value'
WHERE b.columnname IS NULL ;
The other way to update from a select statement:
UPDATE A
SET A.col = A.col,B.col1 = B.col1
FROM first_Table AS A
INNER JOIN second_Table AS B ON A.id = B.id WHERE A.col2 = 'cool'
Option 1: Using Inner Join:
UPDATE
A
SET
A.col1 = B.col1,
A.col2 = B.col2
FROM
Some_Table AS A
INNER JOIN Other_Table AS B
ON A.id = B.id
WHERE
A.col3 = 'cool'
Option 2: Co related Sub query
UPDATE table
SET Col1 = B.Col1,
Col2 = B.Col2
FROM (
SELECT ID, Col1, Col2
FROM other_table) B
WHERE
B.ID = table.ID
UPDATE table1
SET column1 = (SELECT expression1
FROM table2
WHERE conditions)
[WHERE conditions];
The syntax for the UPDATE statement when updating one table with data from another table in SQL Server.
It is important to point out, as others have, that MySQL or MariaDB use a different syntax. Also it supports a very convenient USING syntax (in contrast to T/SQL). Also INNER JOIN is synonymous with JOIN. Therefore the query in the original question would be best implemented in MySQL thusly:
UPDATE
Some_Table AS Table_A
JOIN
Other_Table AS Table_B USING(id)
SET
Table_A.col1 = Table_B.col1,
Table_A.col2 = Table_B.col2
WHERE
Table_A.col3 = 'cool'
I've not seen the a solution to the asked question in the other answers, hence my two cents.
(tested on PHP 7.4.0 MariaDB 10.4.10)