Feat(optimizer): add support for resolving CTEs in CREATE statements #1949
Conversation
| @@ -1006,6 +1006,31 @@ class Create(Expression): | |||
| "clone": False, | |||
| } | |||
|
|
|||
| @property | |||
| def ctes(self): | |||
There was a problem hiding this comment.
should we move these helper methods to a shared base class and have this applied to Insert and Update or other DDLs?
then in the optimizer you can check for the base class instead of just Create
|
thanks so much for the PR! |
|
@gtoonstra are you planning to finish up this pr? |
georgesittas
changed the title
gtoonstra
force-pushed
the
fix_create_insert_not_qualifying
branch
from
4ef83a1 to
c24acaf
Compare
|
@tobymao : I have made the necessary changes and CI is processing now. After this PR, I will look into INSERT/UPDATE as well. |
There was a problem hiding this comment.
LGTM, left one question.
I agree with Toby about renaming TableChange to DDL. I believe INSERT and UPDATE should just work with a few changes (i.e. do instance checks on the common superclass instead of just CREATE).
Btw, if we're to create a new test file for qualify columns, I'd rather use a bit more generic name like qualify_columns_ddl.sql or something to anticipate for the INSERT, UPDATE tests as well.
Nice 👍
| WITH cte AS (SELECT a FROM db.y) CREATE TABLE s AS SELECT * FROM cte; | ||
| WITH cte AS (SELECT a AS a FROM db.y AS y) CREATE TABLE s AS SELECT cte.a AS a FROM cte; |
There was a problem hiding this comment.
Why doesn't cte have an alias in the final SELECT here? Seems a bit off. Ditto for subsequent tests.
|
I'll complete this, but most of the work should be already done. Thanks for the PR @gtoonstra! |
Pull request to resolve CTEs when CREATE statement is at the root of the AST.
I've dived a lot more into the scopes and resolvers and came up with this implementation so far.
Looking forward to feedback.