SQL Server : check if table column exists and remove rows

Question

I'm currently writing a general SQL Server script to cleanup different databases having more/less the same table structure. This script requires that I wipe certain data from a table if that table exists in the database. here a sample of the script

IF EXISTS( SELECT * FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = 'TAB1')
  IF EXISTS( SELECT * FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = 'TAB1' AND  COLUMN_NAME = 'COL1')
    delete TAB1 where COL1 not in (select COL2 from TAB2);

As a programmer, I know that the delete command will not be executed if both condition block are false. However, when I run it in SQL it returns

Invalid column name 'COL1'.

Probably my approach is wrong. Can anyone point me to the right direction?

Answer 1

The problem is, SQL Server wants to compile the entire batch before it executes any of it.

And it can't compile the batch because there's a missing column.

So you have to make sure that the batch can compile without trying to compile the DELETE statement - so keep it as a string and force it to compile separately:

IF EXISTS( SELECT * FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = 'TAB1')
  IF EXISTS( SELECT * FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = 'TAB1'
        AND  COLUMN_NAME = 'COL1')
    EXEC sp_executesql 'delete TAB1 where COL1 not in (select COL2 from TAB2);'

You said:

As a programmer, I know that the delete command will not be executed if both condition block are false.

Assuming, say, a C# background, your original query is like performing two reflection calls to determine whether a type has a particular property, and then having a line of code that directly uses that property on an object of that type - if the type doesn't have the property, the code is not going to compile, so the reflection checks never have a chance to execute.