The latest updates on your projects. Learn more about Vercel for Git ↗︎

Thanks! I think that overall this is the way we should show it and point to the relevant docs but I am just wondering if the example we show is the best one.

Most people will want to upload Python packages so that they can call their favorite functions. In this example we are currently loading a CSV file. It works but it is not going to be the reason why people will want to use the imports config and load it from a Snowflake stage.

Hiya @mirnawong1 and @b-per

Thank you for reviewing this. It's much appreciated.

@b-per with ref to your feedback, I found the below example in the Snowflake doc which gives an example of installing a package with native code via importing from Stage. Would you recommend using this instead:

create or replace function native_module_test_zip()
returns string
language python
runtime_version=3.9
handler='compute'
imports=('@mystage/mycustompackage.zip')
as
$$
import os
import sys
def compute():
   from zipfile import ZipFile
   if os.path.exists('/tmp/mycustompackage'):
      shutil.rmtree('/tmp/mycustompackage')
   os.mkdir('/tmp/mycustompackage')
   zippath = os.path.join(sys._xoptions["snowflake_import_directory"], 'mycustompackage.zip')
   ZipFile(zippath).extractall('/tmp/mycustompackage')
   sys.path.insert(0, '/tmp/mycustompackage')
   #return os.listdir('/tmp/mycustompackage')
   import mycustompackage
   return mycustompackage.mycustompackage()
$$;

Kind Regards
Natalie

Yes, this looks much closer to what people might do! We just need to update it to make it a valid dbt model.

Thanks @b-per

I did a little research and came upon this example:

{{ config(
    materialized = 'view'
) }}

create or replace function native_module_test_zip()
returns string
language python
runtime_version=3.9
handler='compute'
imports=('@mystage/mycustompackage.zip')
as
$$
import os
import sys
from zipfile import ZipFile
import shutil

def compute():
   if os.path.exists('/tmp/mycustompackage'):
      shutil.rmtree('/tmp/mycustompackage')
   os.mkdir('/tmp/mycustompackage')
   zippath = os.path.join(sys._xoptions["snowflake_import_directory"], 'mycustompackage.zip')
   ZipFile(zippath).extractall('/tmp/mycustompackage')
   sys.path.insert(0, '/tmp/mycustompackage')
   
   # Import and call the function from mycustompackage
   import mycustompackage
   return mycustompackage.mycustompackage()
$$;

Would that suffice?

Kind Regards

I don't think so, no. This is not a valid dbt Python model.

There are a few things to change. Our docs describe what a Python model needs to have (for example a def model())

Querying example here: https://dbt-labs.slack.com/archives/C05FWBP9X1U/p1730894409078339

Hiya @mirnawong1 @b-per

I'm currently querying this in ask-adapters-support. I found this example:

def model(dbt, session):
    # Configure the model
    dbt.config(
        materialized="table",
        imports=["@mystage/mycustompackage.zip"],  # Specify the external package location
    )
    
    # Example data transformation using the imported package
    # (Assuming `some_external_package` has a function we can call)
    data = {
        "name": ["Alice", "Bob", "Charlie"],
        "score": [85, 90, 88]
    }
    df = pd.DataFrame(data)

    # Process data with the external package
    df["adjusted_score"] = df["score"].apply(lambda x: some_external_package.adjust_score(x))
    
    # Return the DataFrame as the model output
    return df

Mike Alfare advised: "I think mycustompackage.zip might need to be some_external_package.py. To be honest, this is not something I have seen before, but reviewing our functional tests, it seems like this could work. I'll see if I can find someone more familiar with this particular feature."

I thought I'd share here incase you had any ideas.

Kind Regards
Natalie

I think that both zip and py files would technically work. but people wanting to use external libraries would use the zip approach.

Here are other examples in the internet with the zip option

• https://github.com/sparsh-ai/recohut/blob/21a635dc6a8f16d74632988d580530a9a4023699/docs/03-processing/lab-snowpark-churnpark/models/ml/features/int_user_encoded.py#L45
• https://github.com/phdata/dbt_snowpark_sentiment_example/blob/2c5528278e14dba678fb7773cca2d47f8adbeb4d/models/reviews.py#L30

This snowflake page as well mentions the zip

Rather than waiting for more and more feedback I think that the best is to push the current version and update it later if need be. As it is, it is already providing some guidance on using the imports config which we didn't have before.

Was the snippet completely changed from the one in this thread? The formatting doesn't look as good as the snippet you shared before.

Hiya @b-per

Yep, I used the one you ref'd here. I'll rework it.

Kind Regards