A "negative" join query

Databases
1

Hi,

I have the following problem with the logic for a table join and hope someone here will be able to push me into the right direction.

I have table A with several records and table B with 0-n rows for each record in table A (basically a dynamic list of attributes for each entry in A).
Now I would like to fetch all records from A which do not have a defined list of entries in B (basically “give me all records which do not have all required attributes”).

I believe that the following query would provide me with all matches but as I mentioned I’d be looking for the opposite.

SELECT * FROM A
  LEFT JOIN B ON p1.fk=c.id
  GROUP BY B.attr_name, A.id
  HAVING B.attr_name IN (SELECT 'ATTR1' AS attr UNION ALL SELECT 'ATTR2' UNION ALL SELECT 'ATTR3')

Anyone who can shed a bit of light into this?

2 
SELECT * 
  FROM A
LEFT OUTER
  JOIN B 
    ON B.fk = A.id
   AND B.attr_name IN ( 'ATTR1' , 'ATTR2' , 'ATTR3' )    
GROUP 
    BY A.id
HAVING COUNT(B.attr_name) < 3
3

Thanks Rudy, that appears to work just perfectly! Even though I lost you with the IN() and COUNT() clauses ;).

4

The IN clause replaces a list of ORs

WHERE a=5 OR a=7 OR a=9
can be replaced with
WHERE a IN (5,7,9)

makes for easier typing as your list grows.

The COUNT clause, you were searching for three different attributes and you told us you wanted those that didn’t have all attributes, so therefore the COUNT has to be less than the total number of attributes.

Clear as mud? :slight_smile:

5

Thanks a lot guelphdad, I didnt express myself properly. Actually I am aware of the meaning of IN and COUNT. It just took me a while to figure out the logic of this particular query.

Eventually I realised the key to the task is counting all “attribute” rows for each entry and selecting only those with a count lower than the total. Initially I had a similar approach but without grouping and counting and this led to incorrect results so I hit a wall and lost sight of the forest for the trees :slight_smile:

Thanks again to both of you!