插入案例陈述中

Question

I have a table of data in which I have decided I want to change the way I save and extract data. It was a poor design on my part and now I need to fix it without going through every row manually. I want to change the primary keys from ItemID and VendorID to ItemID and Function. I want a separate record for each function+id rather than a combined record if the vendor id is the same for both functions. I will be eliminating the use of the other_functions column.

Basically my table looks something like this:

ItemID(PK)| VendorID(PK) |   Function   |   other_functions
_______________________________________________________
    1     |    23        |    cooking   |     
    1     |    36        |    cleaning  |
    2     |    45        |    cooking   |   cleaning

After I update the PK's, How do I go through the table, and for each record where other_functions = 'cleaning', insert a new row to make it look more like this:

ItemID(PK)| VendorID     | Function(PK)   
______________________________________
    1     |    23        |    cooking     
    1     |    36        |    cleaning 
    2     |    45        |    cooking
    2     |    45        |    cleaning

I am using PostgreSQL. It seems like it should be a simple solution but haven't been able to find the answer I am looking for. Can this be done with just an insert + case statement?

Answer 1

Can this be done with just an insert + case statement?

CASE would make the value conditional - in your case you want to make the insert itself conditional. A better method would just be to insert records that have a other_functions value:

INSERT INTO {table}
(ItemID, VendorId, Function)
(
SELECT ItemID, VendorId, other_functions
FROM {table}
WHERE other_functions IS NOT NULL
)