Select most recent rows that match a condition in

问题:

Let's say I have a table:

Name, status, timestamp

And I want to select the rows that match status='active' but only those that have the most recent timestamp for each of them. So if there were rows like this:

Bob, active, 5/10/2010
Bob, active, 6/12/2010
Ann, inactive, 6/12/2000
Ann, active, 9/3/2009
Ann, active, 9/25/2010

I'd want it to return:

Bob, active, 6/12/2010
Ann, active, 9/25/2010

How can I do this? I'm using SQLite, if it matters.

Thank you.

回答1:

select name, status, max(timestamp) 
from my_table 
where status = 'active' 
group by name, status

take a look at http://www.w3schools.com/sql/sql_groupby.asp

回答2:

Here's how I solve this type of problem. You want one row for each name such that no other row exists with the same name and a greater timestamp:

SELECT t1.*
FROM MyTable t1
LEFT OUTER JOIN MyTable t2 ON t1.name = t2.name 
  AND t1.timestamp < t2.timestamp
WHERE t2.name IS NULL

But this can still return multiple rows for each name, because you could have more than one row for a given name with the same max timestamp. So use the primary key as a tie-breaker:

SELECT t1.*
FROM MyTable t1
LEFT OUTER JOIN MyTable t2 ON t1.name = t2.name 
  AND (t1.timestamp < t2.timestamp OR t1.timestamp = t2.timestamp AND t1.id < t2.id)
WHERE t2.name IS NULL

I'm assuming id is a primary key for this example, but any other unique column that increases in value chronologically would work.

回答3:

Bill, I believe I have it working now (seems to pick the right rows on different test data). I used your query and added the condition t1.active_status!='active', since that will take care of anything inactive. The whole query looks like:

SELECT t1.*<br>
FROM TrialTypes  t1<br>
LEFT OUTER JOIN TrialTypes t2 ON t1.name = t2.name <br>
  AND (t1.start_date < t2.start_date OR t1.start_date = t2.start_date AND t1.rowid < t2.rowid)<br>
WHERE t2.name IS NULL and t1.active_status != 'active'<br>

Thank you very much for the help. Obviously, I'm new to more than basic SQL queries. This helps teach me, thanks!