Display only needed results in SQL - sql

Question

I need a little assistance in finishing this query. Here is what I have so far:
select
(select count(fileName)
from PDFFile
where dateTime > cast(getdate() as date)
and stateId = 17) AS "Files on SFTP"
,
(select count(fileName)
from PDFFile
where dateTime > cast(getdate() as date)
and stateId = 12) AS "Files Pass"
,
((select count(fileName)
from PDFFile
where dateTime > cast(getdate() as date)
and stateId = 17)
-
(select count(fileName)
from PDFFile
where dateTime > cast(getdate() as date)
and stateId = 12)) AS "Diff"
This is going to give me 3 columns of results. First result will be a number, second will be a number and the third will be the diff. There may even be a better way to write this but I'm still a novice. Hint: There is an entry in the DB for each state:
fileName |dateTime | stateID
--------+---------+-----------------+---------
abc.pdf | 2013-12-17 12:03:14.597 | 17
abc.pdf | 2013-12-17 12:06:23.096 | 12
xyz.pdf | 2013-12-17 12:09:16.583 | 17
xyz.pdf | 2013-12-17 12:10:19.823 | 12
Anyways for the finale...
I need to have a 4th column or a separate query (possible to UNION?) that pulls the fileNames based off the results in the diff.
Hypothetically if the diff is 40, the 4th column or separate query should list the 40 names. At times the diff may be negative so again hypothetically speaking if its -40 it should list the 40 names.
Assistance is greatly appreciated. Thank you!

Answer 1

You can greatly simplify your query using conditional aggregation:
select sum(case when dateTime > cast(getdate() as date) and stateId = 17 then 1 else 0
end) as "Files on SFTP",
sum(case when dateTime > cast(getdate() as date) and stateId = 12 then 1 else 0
end) AS "Files Pass",
(sum(case when dateTime > cast(getdate() as date) and stateId = 17 then 1 else 0
end) -
sum(case when dateTime > cast(getdate() as date) and stateId = 12 then 1 else 0
end)
) as diff
from PDFFile;
To get the list of files that are in the first group but not the second requires a bit more logic. The problem is that the unit of aggregation is at the file level.
select PDFFile
from PDFFile
group by PDFFile
having sum(case when dateTime > cast(getdate() as date) and stateId = 17 then 1 else 0
end) > 0 and
sum(case when dateTime > cast(getdate() as date) and stateId = 12 then 1 else 0
end) = 0;
Each part of the having clause counts the number of rows -- for each file -- that match the two conditions. You want at least one row that matches the first condition (hence > 0) and no rows that match the second (= 0).

Answer 2

This type of "combine row data into one column" question comes up quite a lot on Stack Overflow and although it has its place it's often easier and more efficient to solve the problem in another way.
For example, it's a lot easier to ask SQL to "give me all the filenames where stateid = 17", return them to your app and then get the app to display them. It may also be that your user doesn't want to see them until there is a particular summary line that is of interest to them that they need to drill down into further. Think of email as an example - you may only need to look at the 30 character subject line and know you don't need to download the 1Mb email body.
For your first question though there is a lot easier (and more efficient) way to write your query. Note that this example is untested
select
sum(case when stateId = 17 then 1 else 0 end) as "Files on SFTP",
sum(case when stateId = 12 then 1 else 0 end) as "Files Pass",
sum(case when stateId = 17 then 1 else 0 end) -
sum(case when stateId = 12 then 1 else 0 end) as "Diff",
from
PdfFile
where
datetime > getdate()
I'm using CASE here to prevent having to do three separate sub-queries. Sub-queries are inefficient. CASE isn't great but it's faster than sub-queries. I've also placed your datetime check at the bottom of the query as a WHERE as it was common to each of your checks.