mysql组和过滤结果

Question

I have 2 queries!

select LINE, ACCOUNT
from DATAS
where date between "2015-07-01" and "2015-07-15" 
and SIG="in"
group by ACCOUNT
order by ACCOUNT

and

select LINE, ACCOUNT
from DATAS
where date between "2015-07-01" and "2015-07-15" 
and SIG="out"
group by ACCOUNT
order by ACCOUNT

Sample results:

LINE-ACCOUNT, SIG
0-1309  IN
0-1331  IN
0-1340  IN
0-1361  IN
0-1404  IN
0-1555  IN

LINE-ACCOUNT, SIG
0-1098  OUT
0-1309  OUT
0-1322  OUT
0-1331  OUT
0-1340  OUT
0-1555  OUT

I want 2 queries that will give that results:

1. Give me the ACCOUNTs that have "in" sign, but it hasn't "out" sign!
2. Give me the ACCOUNTs that have "out" sign, but it hasn't "in" sign!

I solved it in excel, but it is too complicated to solve it day by day! ;)

if it is easier I accept the solution in php too!

I hope you understand what I want!

thanks in advance

Answer 1

Have you tried something like this?

select * from 
(
    (select LINE, ACCOUNT, SIG
    from DATAS
    where date between "2015-07-01" and "2015-07-15" 
    and SIG="in"
    group by ACCOUNT
    order by ACCOUNT)
    union all
    (select LINE, ACCOUNT, SIG
    from DATAS
    where date between "2015-07-01" and "2015-07-15" 
    and SIG="out"
    group by ACCOUNT
    order by ACCOUNT)
) t1
group by LINE, ACCOUNT
having count(Line) = 1

EDIT:Corrected the query

EDIT: Here's a brief explanation on how the query works. First off, we select every line that has SIG = 'IN'... then we append every line that has SIG = 'OUT'.

Now is where it gets interesting. If a line has both an IN and an OUT statement, it'll appear twice in our current resultset because it'll be part of both the SIG = "IN" and SIG = "OUT" query. On the other hand, a line that only has an "IN" or an "OUT" will only appear in one of the two inner query. We're going to take advantage of this.

The outer select statement with its "group by LINE, ACCOUNT" statement flattens our resultset and thus allows us to use the count() function. If a line appears in both queries, the count() will equal two... on the other hand, if a line appears only in one of the two queries, the count() will equal one. Those are the lines we want.

Still, in SQL, you can't do:

select * from table where count() > 1

If you want to place a condition using an aggregate function, you have to use the "having" clause, which is placed after a "group by" statement and acts exactly like a "where", only with an aggregate function.