[CT-3017] [Feature] Support show of python model

[ChenyuLInx]

Is this your first time submitting a feature request?

• I have read the expectations for open source contributors
• I have searched the existing issues, and I could not find an existing issue for this feature
• I am requesting a straightforward extension of existing dbt functionality, rather than a Big Idea better suited to a discussion

Describe the feature

dbt-core added show command to help user developing models, however, this command only works for SQL models, when trying to run show selecting a python model, we will get the following error:

03:41:19  Encountered an error:
Runtime Error
  Database Error in model test (models/test.py)
    001003 (42000): SQL compilation error:
    syntax error line 1 at position 0 unexpected 'def'.

This is because dbt-core still tries to run the python model as a SQL.

In order to support this, we likely will have to actually materialize the python model to a temp table, get some results back, and remove the temp table. I am not sure whether the use case here is large enough for dbt-core to support such functionality that is fairly involved.

Describe alternatives you've considered

Not supporting it as python models as sql models works pretty different.

Or we can raise a more clear error saying that show do not support python model

Who will this benefit?

No response

Are you interested in contributing this feature?

No response

Anything else?

No response

[jtcohen6]

Thanks @ChenyuLInx!

First thought: I think this would be easier to support for Snowpark, where we can (in theory) run the Python model as a stored procedure that actually returns rows. I believe that's how it works inside the Snowsight UI. But this doesn't work in the more-general case, whereby Python models are submitted as jobs that produce an object in the DWH.

Second thought: We just need to build the model first — either:

• behind-the-scenes into a temp table (tricky)
• by keeping track of logical state (complex), and rebuilding only when it's changed
• by just exposing this limitation to the user explicitly, and asking them to build the model first after making changes

In any case, this would require that we add a parameter to show to the effect of: "I've already built this model, so please just show me X records from the object that already exists in the DWH"

• [CT-2429] dbt show: option to use already-materialized model instead of rerunning SQL #7391 (comment)

That could be relevant for SQL models, too. It becomes a question of UX and state management: "You haven't changed this model since the last time you built it, so we're going to show you a preview from the existing DWH object rather than re-running the query." Or: "You've edited this Python model since the last you built it, so you need to run it first before previewing."

[github-actions]

This issue has been marked as Stale because it has been open for 180 days with no activity. If you would like the issue to remain open, please comment on the issue or else it will be closed in 7 days.