在Oracle中排除变量的表？

Question

Sorry if this explanation is unclear in any points. I'm working in Oracle.

I have a large table which is structured with columns for Country, Division, SubDivision, SpendCategory, etc. I need to create potentially flexible exclusion criteria for each country for this large table, and was thinking about creating a second 'exclusions table' like so.

 CREATE TABLE EXCLUSIONS

 (COUNTRY VARCHAR(50), EXCLUSIONFIELD VARCHAR(50), FIELDVALUE VARCHAR(50));

My idea was that I could populate this table with rows like

 Mexico - Division - "Central Division"

 Mexico - SubDivision - "Mexico City"

 France - Division - "Paris Division"

...

And with this table, automatically filter my table to exclude data where (country=Mexico AND division="central division"), (country=Mexico AND subdivision="Mexico City"), (country=France AND division="paris division"),

...

So I guess I have two questions. First, is this the best way to do this? I like this idea because the exclusion list is potentially long and having a third party maintain a table seems more feasible than manually updating large numbers of where conditions. Second, how would I write the query to exclude all conditions from the Exclusions table from my main table?

Answer 1

What you are doing can work, but the query will look something like this:

select t.*
from t
where not exists (select 1
                  from exclusions e
                  where e.exclusionfield = 'country' and e.value = t.country or
                        e.exclusionfield = 'district' and e.value = t.district or
                        . . .
                  );

This can be a very hard query to optimize. You might want to express this using separate subqueries:

select t.*
from t
where not exists (select 1
                  from exclusions e
                  where e.exclusionfield = 'country' and e.value = t.country 
                 ) or
      not exists (select 1

from exclusions e where e.exclusionfield = 'district' and e.value = t.district );

You should be able to get reasonable performance with an index on exclusions(value, exclusionfield) .

An alternative would be to replicate the columns . . . country , district and so on, removing the exclusionfield column:

select t.*
from t
where exists (select 1
              from exclusions e
              from e.country = t.country or
                   e.district = t.district or
                   . . .
             );

This might also optimize better with separate exists subqueries.

Answer 2

To exclude data you can use not exist(select 1 from condition_table where ....)
You can also use condition with (country, subdivision) not in(....)
To filter select with some data, you can define table dynamically using 'with' clause before select...