How can I write an UPDATE query for this SELECT query?

I have a query from my previous question (link) abou开发者_开发问答t a complex select, and my question is, how can I write an update query for result of this query, to change fet_id to some new value.

Note that this query will select rows from a queue until rcpts_count reaches 1000, (for traffic controlling). more information about this query is, in this question

   SELECT NULL AS msg_id, NULL AS total, NULL AS found
    FROM dual
     WHERE (
      @total :=0
       OR 
      @found :=0
    )
    UNION
      SELECT msg_id, @total AS total, @found :=1 AS found
        FROM messages_queue
      WHERE (
        @total := @total + rcpts_count
      )
    AND @total <1000
    UNION
      SELECT msg_id, rcpts_count AS total, 0 AS found
        FROM messages_queue
      WHERE IF( @found =0, @found :=1, 0 )

If you're trying to update all the records within the set, you could write a query like this:

UPDATE message_queue mq
INNER JOIN (
   SELECT NULL AS msg_id, NULL AS total, NULL AS found
    FROM dual
     WHERE (
      @total :=0
       OR 
      @found :=0
    )
    UNION
      SELECT msg_id, @total AS total, @found :=1 AS found
        FROM messages_queue
      WHERE (
        @total := @total + rcpts_count
      )
      AND @total <1000
    UNION
      SELECT msg_id, rcpts_count AS total, 0 AS found
      FROM messages_queue
      WHERE IF( @found =0, @found :=1, 0 )
) msgs ON msgs.msg_id = mq.msg_id
SET mq.fet_id = 12345;

However, depending on your activity, this isn't a safe approach, as the records returned from the query could change between the two requests.

I would recommend one of the following approaches:

• Handle the updates on the application side
• Create a temporary table with the result of the query, and update the table with a join to the temporary table
• Assuming 'fet_id' is a key unique to this batch, run the update statement first, then do a simple select based on fet_id.

I see that you are asking for MySql. This is an MsSql solution, but I think the syntax is really close (but not 100% sure of that). This sample is simplified so it can apply to any Stack user. Hopefully, you can translate to your specific dataset.

-- sample table
create table x 
    (col1 int identity(1, 1)
    ,col2 varchar(50))

-- sample data  
insert into x (col2) values
     (null)
    ,(null)
    ,(null)
    ,(null)
    ,(null)

-- update from select
update x
    set x.col2 = 'even'
from x as [t2]
where 
    col1 = t2.col1
    and t2.col1 % 2 = 0

-- show results
select * from x

-- clean up
drop table x

Best of luck.