加入或子查询以“检测丢失的记录”？

Question

I'm struggling to get this query working.

I have 2 tables that contain a countrycode being 'US' or 'AU' or 'JP' - this is a common key.

SELECT DISTINCT cc FROM geo_world; SELECT cc FROM geo_country;

Note: geo_world is a large table with millions of entries but only 124 distinct cc values.

Note: geo_country contains only 244 cc values.

I want to find out the cc values in geo_country that are not show in geo_world. There should be 120 of them.

thankyou - I've tried joins but failed to get the result.

Answer 1

In principle, any of these should work:

SELECT cc
  FROM geo_country
 WHERE cc NOT IN
        ( SELECT cc                 -- or SELECT DISTINCT cc
            FROM geo_world
        )
;

SELECT cc
  FROM geo_country
 WHERE NOT EXISTS
        ( SELECT 1
            FROM geo_world
           WHERE cc = geo_country.cc
        )
;

SELECT geo_country.cc
  FROM geo_country
  LEFT
 OUTER
  JOIN geo_world
    ON geo_world.cc = geo_country.cc
 WHERE geo_world.cc IS NULL             -- i.e., the join failed
;

But you'll have to try them to see which one performs best.

Answer 2

SELECT DISTINCT cc
FROM geo_world
WHERE cc NOT IN (SELECT cc FROM geo_country)

Some notes:

1. Make sure that geo_world has an index that covers cc
2. If you need to do that on regular basis - you probably better denormalize the data and keep unused countrycodes in another small 100-or-some rows table, just as some sort of cache

Answer 3

You can use NOT EXISTS condition. You can do this like this:

SELECT cc FROM geo_country WHERE 
NOT EXISTS (SELECT * from geo_world WHERE geo_world.cc = geo_country.cc);