Snowflake Python Demos API¶

Install the Snowflake Demos library¶

You can install the Snowflake Demos library for use with conda or a virtual environment. To set up the library, follow these steps:

1. Activate a Python environment.

2. To install the library, run the following pip install command:

   pip install snowflake.demos
   

   Copy

Set up a default Snowflake connection¶

The Snowflake Demos API uses the default connection for the Snowflake Python Connector. To configure this connection, follow the instructions in Setting a default connection.

For example, to specify a named connection as the default connection in your Snowflake config.toml file, you add your default connection name to the config.toml file as follows:

default_connection_name = '<connection_name>'

For information about specifying connection definitions in a TOML configuration file, see Connecting using the connections.toml file.

Import snowflake.demos¶

To use the library in your terminal, you can open an interactive shell such as the standard Python REPL.

1. Run the following command (which might vary depending on your Python environment):

   python3
   

   Copy

2. In the REPL session, to import the library and the relevant functions, run the following code:

   from snowflake.demos import help, load_demo, teardown
   

   Copy

Current list of available demos¶

The Snowflake Demos API currently works with the following list of demos:

Load and explore a demo¶

• To load a specific demo and set up its associated resources in Snowflake, call load_demo() with an argument that specifies the demo_name of any available demo, as found in the help() output.

  For example:

  load_demo('get-started-snowflake-ml')
  

  Copy

This function does the following:

• Creates a connection to Snowflake if it’s the first time you’re loading a demo.

• Creates the necessary notebooks.

• Displays the notebook URL for the first step of the demo (step 1), if you are not assigning load_demo() to a variable.

  • If you assign load_demo() to a variable, you need to call demo.show() to get the first notebook URL.

The output should look similar to the following:

Connecting to Snowflake...✅
Using ACCOUNTADMIN role...✅
Creating Database SNOWFLAKE_DEMO_DB...✅
Creating Schema SNOWFLAKE_DEMO_SCHEMA...✅
Creating Warehouse SNOWFLAKE_DEMO_WH...✅
Creating Stage SNOWFLAKE_DEMO_STAGE...✅
Uploading files to stage SNOWFLAKE_DEMO_STAGE/get-started-snowflake-ml and creating notebooks...
Creating notebook get_started_snowflake_ml_start_here...✅
Creating notebook get_started_snowflake_ml_sf_nb_snowflake_ml_feature_transformations...✅
Creating notebook get_started_snowflake_ml_sf_nb_snowflake_ml_model_training_inference...✅
Creating notebook get_started_snowflake_ml_sf_nb_snowpark_ml_adv_mlops...✅
Running setup for this demo...✅

View the demo URL¶

You can use the show() function to view the URL to a specific step in the demo.

• To view the URL for the current step, first assign the result of load_demo() to a variable, such as demo, and then call show() with no arguments:

  demo.show()
  

  Copy

  The output should look similar to this:

  Showing step 1.
  Please copy and paste this url in your web browser to open the notebook:
  https://app.snowflake.com/myorg/myaccount/#/notebooks/SNOWFLAKE_DEMO_DB.SNOWFLAKE_DEMO_SCHEMA.GET_STARTED_SNOWFLAKE_ML_START_HERE
  

• To get the notebook URL for a specific step in the demo, pass the step argument with a specified step number to show():

  demo.show(step=1)
  

  Copy

• To get the notebook URL for the next step in a multi-step demo, use the show_next() function:

  demo.show_next()
  

  Copy

Delete a demo and its resources¶

When you’re done exploring the demos that you set up, you might want to clean up all created resources, datasets, and notebooks that were created.

• To delete a single demo and its associated resources, first assign the result of load_demo() to a variable such as demo, and then call teardown() on it:

  demo.teardown()
  

  Copy

• To delete all demos and any associated resources that have been set up, call teardown() as a top-level function:

  teardown()
  

  Copy