Hey I m having an issue linking the output of a newly create

Question

Hey I m having an issue linking the output of a newly created snowflake table to pandas All the code below works however I want to find a solution that avoids creating a `my new table` source asset It works as follows 1 production cement creates a new table in snowflake called PRODUCTION 2 PRODUCTION is imported as a new SourceAsset 3 my derived asset reads the new SourceAsset into pandas Is there a way of avoiding step 2 and loading the output of 1 into 3 ```ngfs = SourceAsset key= NGFS asset required resource keys= snow resource non argument deps= NGFS def production cement context gt None Create a new table on Snowflake called PRODUCTION context resources snow resource execute query create or replace view PRODUCTION as select from NGFS where variable = Production|Cement my new table = SourceAsset key= PRODUCTION asset io manager key= pd snowflake manager ins= upstream AssetIn PRODUCTION def my derived asset upstream pd DataFrame return upstream dropna ```

sandy · Answer

Hi < Phillip Marks> have you tried setting `io manager key= pd snowflake manager ` on the `production cement` asset

Phillip Marks · Answer

Hey < sandy> switching to `pd snowflake manger` rather then `snowflake resource` I get the error AttributeError DbIOManager object has no attribute execute query using `context resources pd snowflake manager execute query SQL ` does pd snowflake manger have an equivalent to `snowflake resource` s `execute query`

sandy · Answer

I didn t mean to suggest that you should replace the `snowflake resource` with the `pd snowflake manger` But rather that you should keep doing what you re doing and also set the `io manager key`

Phillip Marks · Answer

is the idea that production cement outputs a pandas df If so gt None works but doesn t output a pd Dataframe ``` asset required resource keys= snow resource non argument deps= NGFS io manager key= io manager def production cement context gt None context resources snow resource execute query create or replace view PRODUCTION as select from NGFS where variable = Production|Cement ``` this ``` def production cement context gt pd DataFrame ``` and this ``` def production cement context pd DataFrame ``` don t or can I modify `my derived asset` to accept `production cement` as a pd Dataframe

sandy · Answer

gt is the idea that production cement outputs a pandas df If so gt None works but doesn t output a pd Dataframe it doesn t need to output a pandas DF putting the IO manager on the asset just tells downstream assets that depend on it how to load it

Phillip Marks · Answer

righto how would I change the decorator of `my derived asset` to be downstream of production cement ``` asset io manager key= pd snowflake manager ins= upstream AssetIn PRODUCTION def my derived asset upstream pd DataFrame return upstream dropna ```

sandy · Answer

``` asset io manager key= pd snowflake manager def my derived asset production cement pd DataFrame return production cement dropna ```

Phillip Marks · Answer

thanks I tried that but get the error DATABASE SCHEMA PRODUCTION CEMENT does not exist because production cement creates table PRODUCTION how can I pass PRODUCTION to my derived asset

sandy · Answer

so you want the table to have a different name than the asset right now the built in snowflake pandas IO manager uses the asset name as the table name so you d need to either change the name of the asset or write an IO manager that does the translation

Phillip Marks · Answer

thank you so much for your help < sandy>