MySQL Union Operator not behaving as expected - mysql

Question

I usually find the answer to my questions here. If not, I know that is probably because I haven't looked deep enough. However this time none of the answers I looked through fit my needs. So here it goes:
I want to use MySQL server to build a report from two tables:
ID Email ID Date Spent
----------------------- ---------------------------
A123 a#test.com A123 3.3.14 2.50
B102 b#test.com A123 7.3.14 3.50
yum yum#a.com B102 4.4.14 7.00
(null) (null)
I want to make a report in which for a given timestamp, eg. from 3.1.2014 to 3.31.14, I get a list of all the ID's of the system with the corresponding amount that they have sent, even if they didn't spend anything.
Therefore from the above tables, let's say we want to retrieve the month of March 2014 (from 3.1.2014 to 3.31.14), I would like to get:
ID Spent
------------------
A123 6.00
B102 7.00
yum NULL
I don't mind to have NULL or 0's for those users that didn't spend any money. So far I have gotten this:
SELECT ID,
sum(Spent) AS Spent
FROM expenses
WHERE expenses.date >= '2014-03-01 00:00:00'
AND expenses.date < '2014-03-31 00:00:00'
GROUP BY expenses.ID
UNION
SELECT users.ID,
NULL AS Spent
FROM users
WHERE users.ID NOT IN
(SELECT expenses.ID
FROM expenses
WHERE expenses.date >= '2014-03-01 00:00:00'
AND expenses.date < '2014-03-31 00:00:00'
GROUP BY expenses.ID)
ORDER BY ID;
You can check the above code in here.
It works as expected, but in the real case of my database, the amount of rows in the result query is greater that the amount of unique rows in users.
I have checked that all the ID's in the expenses table are in the users tables, which is the case. I have tested with:
SELECT ID FROM expenses
WHERE expenses.date >= '2014-03-01 00:00:00'
AND expenses.date < '2014-03-31 00:00:00'
AND ID NOT IN (SELECT users.ID FROM users);
and it returns an empty set, which is as expected.
I must be missing something, but I have no clue of what. Could someone please give some insights? I am pretty new to MySQL, and maybe there is a better way of doing this.

Answer 1

I think you want a left outer join rather than a union:
SELECT u.ID, sum(e.Spent) AS Spent
FROM users u left outer join
expenses e
on u.id = e.id and
e.date >= '2014-03-01 00:00:00' and
e.date < '2014-03-31 00:00:00'
GROUP BY u.ID;
Note in the data you have in the SQL Fiddle, this will return four rows, because of the NULL valued row in users.

Answer 2

you can do a join of two tables (Users and Expenses) to get the same result (http://sqlfiddle.com/#!2/738ea6/5). This will be much faster and an understandable query.
SELECT expenses.ID,
sum(Spent) AS Spent
FROM expenses, users
WHERE expenses.date >= '2014-03-01 00:00:00'
AND expenses.date < '2014-03-31 00:00:00'
AND users.ID = expenses.ID
GROUP BY expenses.ID
ORDER BY ID;

Answer 3

Group by in second query?
SELECT ID,
sum(Spent) AS Spent
FROM expenses
WHERE expenses.date >= '2014-03-01 00:00:00'
AND expenses.date < '2014-03-31 00:00:00'
GROUP BY expenses.ID
UNION
SELECT users.ID,
NULL AS Spent
FROM users
WHERE users.ID NOT IN
(SELECT expenses.ID
FROM expenses
WHERE expenses.date >= '2014-03-01 00:00:00'
AND expenses.date < '2014-03-31 00:00:00'
GROUP BY expenses.ID)
GROUP BY ID
ORDER BY ID;