How to execute commands/functions in order - sql

Question

I'm writing a long SQL query that I will be using to automate the process of ingesting large-ish flat files (using python to flatten heavily nested JSON files) and normalizing them for scalability and ease of use with PowerBI reports and dashboards.
Currently, I've got a long process that slices the table into multiple tables, generates mapping tables between them and the primary table, remaps a PK/FK link back to the primary table and drops the old unneeded columns from the primary table.
I'm still building and debugging the script, and I'm getting really frustrated with something that I think I'm doing wrong as I'm not very proficient in SQL.
Currently, if I try to run all of my code at once it will fail saying I'm using invalid column names. The column names are invalid with the tables in their current state, but if it would simply execute from top to bottom, they would be valid by the time it got to them. I've got to highlight and execute my drop tables statement by itself every time I want to rerun the entire script even though I've got the same drop tables statement at the top.
Any advice on how to make the script simply execute from top to bottom or how to make it step through and ignore the "current" state of the tables (prior to execution) would be greatly helpful.
Some example pseudo of what I've got:
CREATE OR ALTER PROCEDURE DropTables
AS
BEGIN
DROP TABLE IF EXISTS
t1,
t2,
t3
END
GO
CREATE OR ALTER PROCEDURE GenerateTable1
AS
BEGIN
~make table~
END
GO
CREATE OR ALTER PROCEDURE GenerateTable2
BEGIN
~make table~
END
GO
CREATE OR ALTER PROCEDURE GenerateTable3
BEGIN
~make table~
ALTER TABLE t1 ADD ~fk from t3~
UPDATE t1
SET ~keys to match~
FROM t3 WHERE t1.old_col = t3.new_col
ALTER TABLE t1
DROP COLUMN old_col
END
GO
EXEC DropTables
GO
EXEC GenerateTable1
GO
EXEC GenerateTable2
GO
EXEC GenerateTable3
Upon executing this I get "Invalid column name old_col" because old_col currently doesnt exist, however, if it would just execute from top to bottom, old_col would exist when it got to it.
Current workaround is highlighting droptables and executing it by itself first, then I can execute everything at once

Answer 1

GO breaks the script into batches. You just need to scope the batches so each one compiles. Or use dynamic SQL which is just a different way to issue separate batches.

Answer 2

It may not be the most elegant solution, but simply wrapping everything in an exec block with single quotes seems to work. "EXEC(' stuff ');" SQL doesnt try to get ahead of itself and is forced to execute in order from top to bottom, example:
CREATE OR ALTER PROCEDURE DropTables
AS
BEGIN
EXEC('
DROP TABLE IF EXISTS
t1,
t2,
t3
')
;
END
GO
CREATE OR ALTER PROCEDURE GenerateTable1
AS
BEGIN
EXEC('
~make table~
')
;
END
GO
CREATE OR ALTER PROCEDURE GenerateTable2
BEGIN
EXEC('
~make table~
')
;
END
GO
CREATE OR ALTER PROCEDURE GenerateTable3
BEGIN
EXEC('
~make table~
ALTER TABLE t1 ADD ~fk from t3~
UPDATE t1
SET ~keys to match~
FROM t3 WHERE t1.old_col = t3.new_col
ALTER TABLE t1
DROP COLUMN old_col
')
;
END
GO
EXEC DropTables
GO
EXEC GenerateTable1
GO
EXEC GenerateTable2
GO
EXEC GenerateTable3