Question

Say we have a table like this:

**tablename**
ID     Name      Size      Date   AtStore1  AtStore2
 1    Apple    Medium  20120101                  Yes
 2     Pear    Medium  20111231        Yes       Yes
 3    Lemon     Small  20111231        Yes       Yes
 4   Orange     Small  20111231        Yes          
 5   Carrot    Medium  20111231        Yes          
 6   Potato     Small  20111231        Yes          
 7   Celery     Large  20111231                  Yes
 8    Onion    Medium  20111231                     
 9   Tomato    Medium  20111231                     
 10   Apple    Medium  20111231                     
 12    Pear    Medium  20111230        Yes       Yes
 13   Lemon     Small  20111230        Yes       
 14  Orange     Small  20111230                  Yes
 15  Carrot    Medium  20111230                  Yes
 16  Potato     Small  20111230           
 17  Celery     Large  20111229                     
 18   Onion    Medium  20111229                  Yes
 19  Tomato    Medium  20111229                     

Can we construct an efficient query that retrieves ALL data per row where:

• “Yes” AtStore1 rows where there are at least 4 “Yes”es in AtStore1 on that day single day
  OR (inclusive)
• There is a “Yes” in AtStore2

Also acceptable is if only the first parameter is satisfied, i.e. the forward lookup of rows. I can probably script the AtStore2 part with PHP if necessary. All of my attempts have failed miserably; my attempts on how to write the question effectively in Google weren’t fruitful either.

(Answers breaking things out into different parts of code in PHP are fine too, I just want something reasonably efficient.)

For this example table, these would be the expected rows:

ID     Name      Size      Date   AtStore1  AtStore2
 1    Apple    Medium  20120101                  Yes
 2     Pear    Medium  20111231        Yes       Yes
 3    Lemon     Small  20111231        Yes       Yes
 4   Orange     Small  20111231        Yes          
 5   Carrot    Medium  20111231        Yes          
 6   Potato     Small  20111231        Yes          
 7   Celery     Large  20111231                  Yes
12     Pear    Medium  20111230        Yes       Yes
14   Orange     Small  20111230                  Yes
15   Carrot    Medium  20111230                  Yes
18    Onion    Medium  20111229                  Yes

As you can see

• Jan 1 2012
  -No Yeses in AtStore1
  -AtStore2 has a Yes so the row is returned
• Dec 31
  -5 Yeses in AtStore1, so those are returned (AtStore2 Yeses in ID 2 and 3 would have sufficed as well)
  -Also has ID 7 since there is a Yes in AtStore2
• The rest of the rows returned were only because of Yeses in AtStore2

Answer 1

Try this:

select t.* 
from tablename t
join (
    select date
    from tablename
    where AtStore1='yes'
    group by date
    having count(*) >= 4
) ta on t.date = ta.date and t.AtStore1 = 'yes'
union
select *
from tablename
where AtStore2 = 'yes'

Result:

ID  Name    Size    Date    AtStore1    AtStore2
1   Apple   Medium  2012-01-01 00:00:00.000 NULL    Yes
2   Pear    Medium  2011-12-31 00:00:00.000 Yes Yes
3   Lemon   Small   2011-12-31 00:00:00.000 Yes Yes
4   Orange  Small   2011-12-31 00:00:00.000 Yes NULL
5   Carrot  Medium  2011-12-31 00:00:00.000 Yes NULL
6   Potato  Small   2011-12-31 00:00:00.000 Yes NULL
7   Celery  Large   2011-12-31 00:00:00.000 NULL    Yes
12  Pear    Medium  2011-12-30 00:00:00.000 Yes Yes
14  Orange  Small   2011-12-30 00:00:00.000 NULL    Yes
15  Carrot  Medium  2011-12-30 00:00:00.000 NULL    Yes
18  Onion   Medium  2011-12-29 00:00:00.000 NULL    Yes

In this case null values are intentionally eliminated by an aggregate.

I’m not sure about and t.AtStore1 = 'yes' in ta on t.date = ta.date and t.AtStore1 = 'yes' remove it if you want all rows from specific day without caring whether it has yes value in AtStore1 column.

Added:

Answer to question from comments, but this is quick and probably dirt way to do it (lack of time):

select t3.*, t4.qty from tablename t3
join (
    select id, max(qty) as qty from (
        select t.id, ta.qty
        from tablename t
        join (
            select date, count(*) as qty
            from tablename
            where AtStore1='yes'
            group by date
            having count(*) >= 4
        ) ta on t.date = ta.date and t.AtStore1 = 'yes'
        union
        select id, 0 as src
        from tablename
        where AtStore2 = 'yes'
    ) t2
    group by t2.id
) t4 on t3.id = t4.id

Result:

ID  Name    Size    Date    AtStore1    AtStore2    qty
1   Apple   Medium  2012-01-01 00:00:00.000 NULL    Yes 0
2   Pear    Medium  2011-12-31 00:00:00.000 Yes Yes 5
3   Lemon   Small   2011-12-31 00:00:00.000 Yes Yes 5
4   Orange  Small   2011-12-31 00:00:00.000 Yes NULL    5
5   Carrot  Medium  2011-12-31 00:00:00.000 Yes NULL    5
6   Potato  Small   2011-12-31 00:00:00.000 Yes NULL    5
7   Celery  Large   2011-12-31 00:00:00.000 NULL    Yes 0
12  Pear    Medium  2011-12-30 00:00:00.000 Yes Yes 0
14  Orange  Small   2011-12-30 00:00:00.000 NULL    Yes 0
15  Carrot  Medium  2011-12-30 00:00:00.000 NULL    Yes 0
18  Onion   Medium  2011-12-29 00:00:00.000 NULL    Yes 0