Use statement #936
Use statement #936
Conversation
|
@xzdandy @gaurav274 You can give read about the design. I will add more implementations to check in binder/catalog. |
|
For use query, we have the option to skip the binder and optimizer using SKIP_BINDER_AND_OPTIMIZER_STATEMENTS. This can simplify the implementation. |
I meant just checking if the data source is valid. Not gonna check anything for the native query. |
| @@ -172,11 +174,19 @@ explain_statement: EXPLAIN explainable_statement | |||
|
|
|||
| explainable_statement : select_statement | insert_statement | update_statement | delete_statement | create_table | |||
|
|
|||
| // Context Statements | |||
|
|
|||
| use_statement: USE database_name "{" query_string "}" // One shortcoming that query string cannot have parenthesis | |||
There was a problem hiding this comment.
Is this the limitation of the lark or ? Why we can not have parenthesis?
There was a problem hiding this comment.
I will remove this comment. The native query can have parenthesis after I change to use curly bracket if that makes sense
// now
USE postgres {
// some query
}
// before
USE postgres (
// some query
)If I use the parenthesis and the native query has arbitrary parentheses as well, I have some trouble of coming up with a grammar that works for all cases.
evadb/optimizer/operators.py
Outdated
| @@ -1239,3 +1240,37 @@ def __hash__(self) -> int: | |||
| self.search_query_expr, | |||
| ) | |||
| ) | |||
|
|
|||
|
|
|||
| class LogicalUse(Operator): | |||
There was a problem hiding this comment.
Can you also bypass optimizer for use? Just to simplify the code
There was a problem hiding this comment.
Will push the code to take care of this
| engine = create_engine(conn_str) | ||
|
|
||
| with engine.connect() as con: | ||
| if "SELECT" in self._query_string or "select" in self._query_string: |
There was a problem hiding this comment.
We need to make this robust for cases where select is used for creating tables etc. Also, if query contains Select etc
There was a problem hiding this comment.
What is the output for CREATE TABLE from sqlalchemy? Is it possible to simply execute the query and yield whatever output sqlalchemy yields in dataframe type?
There was a problem hiding this comment.
How about we replicate this? https://github.com/mindsdb/mindsdb/blame/b417ecff78cca2e7f09d4e7daa3461595c77ad0d/mindsdb/integrations/handlers/postgres_handler/postgres_handler.py
I'm using it in the other PR for testing the connection.
There was a problem hiding this comment.
sqlalchemy/sqlalchemy#5433 (comment)
Based on what I see, it is still true for this API for latest SQLAlchemy. There is no rows returned besides SELECT.
Working example
USE postgres { CREATE TABLE (...) }