SQL NOT EXIST 子查询

Question

I have a situation when I have a number of unique site/primary records with child records, and subsequent grandchild records. I am looking to modify my SQL query so it will populate the unique primary records based on whether or not grandchild records exist, but then further filter down those primary records based on the child record columns.

Eg

SELECT PrimaryTable.SiteKey
FROM PrimaryTable
WHERE NOT EXISTS (
    SELECT*
    FROM PrimaryTable, ChildTable, GrandChildTable
    WHERE PrimaryTable.SiteKey = ChildTable.SiteKey
    and Childtable.Key = GrandChildTable.ParentKey)

This works, whereby the list of sites/primary records that have child records, which do not have grandchild records populate.

If I want to further filter on the Primary table, to the end of the query I could add AND PrimaryTable.Column1 = 'MyDesiredValue' and it works fine. However, I want to filter it by ChildTable.Column != 'MyDesiredValue' .

When I do this, I get an error that the parameter is not bound, and the ChildTable is not listed at the beginning of the query.

I need the Primary Keys to populate and I cannot use joins using the interface/software I am working with, unfortunately. Hoping there is a way to modify the query above to make it work.

Answer 1

FIrst, convert your query to use an outer (ie LEFT) join, that filters out rows with no grand children. Then add your filter to the join:

SELECT distinct PrimaryTable.SiteKey -- distinct, or it'll repeat every join
FROM PrimaryTable
JOIN ChildTable on PrimaryTable.SiteKey = ChildTable.SiteKey -- inner (required) join
    AND ChildTable.Column != 'MyDesiredValue'
LEFT JOIN GrandChildTable on Childtable.Key = GrandChildTable.ParentKey
    AND GrandChildTable.year = 2020
WHERE GrandChildTable.ParentKey IS NULL -- only rows without grandchildren