DB2 SQL更新中的案例结构

Question

I need to update a value (FIELD_E) in one table based on the value in a field of a second table as long as certain conditions are met. Here is what I am working with thus far.

TABLE_1.FIELD_C is a primary key

TABLE_2.FIELD_D is a foreign key

There is a ONE TO MANY relationship between TABLE_1.FIELD_C AND TABLE_2.FIELD_D

UPDATE TABLE_1
SET    TABLE_1.FIELD_E = (
   CASE
          WHEN
                 (
                        SELECT MAX(FIELD_A)
                        FROM   TABLE_1,
                               TABLE_2
                        WHERE  TABLE_1.FIELD_C = TABLE_2.FIELD_D
                        AND    TABLE_1.FIELD_A IS NOT NULL
                        AND    TABLE_2.FIELD_B IS NULL) = 'Y' THEN '1'
          WHEN
                 (
                        SELECT MAX(FIELD_A)
                        FROM   TABLE_1,
                               TABLE_2
                        WHERE  TABLE_1.FIELD_C = TABLE_2.FIELD_D
                        AND    TABLE_1.FIELD_A IS NOT NULL
                        AND    TABLE_2.FIELD_B IS NULL) = 'N' THEN '10'
          WHEN
                 (
                        SELECT MAX(FIELD_A)
                        FROM   TABLE_1,
                               TABLE_2
                        WHERE  TABLE_1.FIELD_C = TABLE_2.FIELD_D
                        AND    TABLE_1.FIELD_A IS NOT NULL
                        AND    TABLE_2.FIELD_B IS NULL) = 'U' THEN '9'
          ELSE NULL
   END 
WHERE TABLE_1.FIELD_C = TABLE_2.FIELD_D
AND    TABLE_1.FIELD_A IS NOT NULL
AND    TABLE_2.FIELD_B IS NULL

I know I could probably do this more simply by just taking a snapshot and writing multiple statements using primary keys but if the values were to change after the snapshot and before the update, an incorrect update could be made.

Can I get some suggestions on how to do this using a similar case structures statement?

Answer 1

Your query is doing several things you likely do not intend, besides the blatant syntax and construction errors. Making some assumptions about reasonable behavior, the query can be a lot simpler than your current effort:

UPDATE Table_1 SET field_e = CASE field_a WHEN 'Y' THEN '1'
                                          WHEN 'N' THEN '10'
                                          WHEN 'U' THEN '9'
                                          ELSE null END
WHERE field_a IS NOT NULL
      AND EXISTS (SELECT '1'
                  FROM Table_2
                  WHERE Table_2.field_d = Table_1.field_c
                        AND Table_2.field_b IS NULL)

SQL Fiddle Example

---

Your original query had the following behavior or other problems:

1. Your subqueries were certainly returning more than one row (error)
2. Your outermost SELECT mentions Table_2 without providing a reference (error)

3. There was no correlation between the outer UPDATE and any of the subqueries, meaning everything would be updated to have 1 , regardless of the actual contents of a particular row's field.

   • I'm also not sure if you want the ELSE NULL in your CASE - this has the effect of erasing whatever value is currently in field_e if field_a doesn't have a matching value. Or rather, not restricting the update to those rows containing the values interested in (because ELSE null is implied if excluded).