Reputation: 1434
I need help with some MySQL pain in the ****... Anyway, i got the following sql :
SELECT id,count(*),
CASE
WHEN count(*) > 1 THEN
// I need the minimal `taskdate_time` column from the selected rows
// where a certain boolean is active
ELSE taskdate_time
END
FROM timehistory th
WHERE `send`=true
GROUP BY date_format(taskdate_time, "%Y-%m-%d"), user_id
As described in the comments, i need to get the earliest time out for the two rows where a column called removed
is not FALSE
How do i achieve this?
My columns are :
`id` - int
`taskdateuser_id` int
`user_id` int
`changed_by` int
`batch_id` int
`taskdate_time` timestamp
`send` tinyint
`isread` tinyint
`update` tinyint
`removed` tinyint
Many thanks in advance!!!
EDIT:
I might explain it a bit more. If i got the following table rows :
The red marked rows are captured by the CASE count(*) > 1, because there are 2 rows returned by the group by. Then i need to to a SELECT from that 2 captured rows where removed=false
and min(taskdate_time)
. So if 4 rows are returned for that group by, and 2 of the rows are removed=false
and the other are removed=true
then i need to do a subselect for the minimum taskdate_time
that 2 rows where removed=false
.
Upvotes: 0
Views: 65
Reputation: 62851
You could try something like this:
SELECT TH.user_id, COUNT(*),
CASE WHEN COUNT(*) > 1
THEN MIN(IF(TH.removed, TH.taskdate_time, NULL))
ELSE TH.taskdate_time
END
FROM TimeHistory TH
...
However, if COUNT > 1 AND there aren't any records where TH.removed is true, then this will return NULL for that value. What should it return in those cases?
--EDIT--
In response to comments, then this should work just wrapping it with COALESCE
:
COALESCE(
CASE
WHEN COUNT(*) > 1
THEN MIN(IF(TH.removed, TH.taskdate_time, NULL))
ELSE TH.taskdate_time
END, MIN(TH.taskdate_time))
Upvotes: 0
Reputation: 263813
SELECT id,
count(*),
CASE WHEN count(*) > 1
THEN (SELECT MAX(taskdate_time) FROM timehistory f WHERE f.id = th.id AND removed = 0)
ELSE taskdate_time
END
FROM timehistory th
WHERE `send` = true
GROUP BY date_format(taskdate_time, "%Y-%m-%d"), user_id
Upvotes: 1