Upload and run custom functions in a clean room¶

Running uploaded code¶

Each bundle of uploaded code can define multiple functions that call each other, but a bundle exposes only one handler function. This handler function can be called by templates created or run by anyone who uses the clean room. If the code creates internal tables, these tables can be accessed as described in Designing multi-step flows.

For example, if you uploaded a function named simple_add that takes in two numeric parameters, you can call it from a template as shown here. The function is always referenced using the scope cleanroom. For example, a template could call simple_add like this:

SELECT cleanroom.simple_add({{ price | sqlsafe | int }}, {{ tax | sqlsafe | int }}) ...

SELECT
  cleanroom.simple_add(
    {{ price | sqlsafe | int }}, {{ tax | sqlsafe | int }}
    ) AS TOTAL_ITEM_COST
...

You can upload multiple functions in a single package, and functions within a single package can call each other, but functions can’t call functions within other packages. (They can call the handler functions, though.) For example:

Notes:

• Code uploaded by either side (provider and consumer) can be run by either side.

• A template can call function A or function B, but not A1, A2, B1, B2.

• Function A can call function B, and the reverse.

• Function A can’t call B1 or B2 and Function B can’t call A1 or A2.

• A1 can call A2 and the reverse. A1 and A2 can call B. A1 and A2 can’t call B1 or B2.

• B1 can call B2 and the reverse. B1 and B2 can call A. B1 and B2 can’t call A1 or A2.

Updating or deleting custom functions¶

You can upload or overwrite an existing function or template that you uploaded, but you can’t delete an existing function or template. The only way to “remove” a function is to create a dummy function with the exact same name and signature that always succeeds.

Uploading a function with the exact same signature as one that you previously uploaded will overwrite the existing function. The signature is the case-insensitive function name of the external handler, and the data types of all parameters, in the same order. Parameter names don’t matter. You can’t overwrite a function uploaded by another account.

Because the signature must match when you update a function, you cannot change the signature of an existing function: if you upload the function foo(name VARIANT age INTEGER) and then upload the function foo(name VARIANT age FLOAT), the second function will be added to the clean room in addition to the first, because the argument types differ.

Important notes about versioning¶

Every time the provider uploads a function, it increases the patch number (and there is a limit of 99 patch numbers). Therefore, do your best to test and debug your code thoroughly before adding it to the clean room to reduce version updates during development.

If you do update a patch number, customers using the clean room UI might need to refresh the page to see the change. Customers using the API should see changes immediately, but there can be a delay, depending on the available resources. Learn more about clean room versioning.

Uploading provider-written inline functions¶

You can upload the code inline in the code parameter of provider.load_python_into_cleanroom. Here is an example of uploading a simple function inline:

CALL samooha_by_snowflake_local_db.provider.load_python_into_cleanroom(
$cleanroom_name,
'simple_add',                         -- Name used to call the UDF from a template.
['first INTEGER', 'second INTEGER'],  -- Arguments of the UDF, specified as '<variable_name> <SQL type>' pairs.
['numpy', 'pandas'],                  -- Packages imported by the UDF.
'INTEGER',                            -- SQL return type of UDF.
'add_two',                            -- Handler function in your code called when external name is called.
$$
import numpy as np   # Not used, but you can load supported packages.
import pandas as pd

def add_two(first, second):
    return first + second
$$
);

The calling template calls cleanroom.simple_add to call this function. The provider examples demonstrate how to upload inline code.

Uploading provider-written functions from a stage¶

You can upload Python files to a clean room stage and reference the stage when you call provider.load_python_into_cleanroom. Loading code from a stage allows you to develop the code in your local system in an editor, avoid copy/paste errors when loading it inline, and also have better versioning control. Note that you can upload multiple files in a single procedure call, but only one handler function is exposed for each upload.

Code is loaded from a stage into the clean room when you call load_python_into_cleanroom; later changes to the code on the stage are not propagated to the clean room.

To upload your UDF to a stage:

1. Create your .py file and make it available in a location where you can upload it to a Snowsight stage.

2. To get the name of the stage for your clean room, call provider.get_stage_for_python_files($cleanroom_name). This stage is accessible by the clean room – you cannot use an arbitrary stage that you create.

3. Upload the .py file to the stage for your clean room. There are several ways to do this, including using the CLI, Snowsight, or language-specific drivers.

4. Call provider.load_python_into_cleanroom with the stage location, handler, external name, arguments, and return type. Templates in your clean room can now call the function.

The following example code shows how to load code into a clean room from a stage.

-- Save the following code as reverser.py:
--import numpy as np
--def main(some_string):
--  '''Return the reverse of a string plus a random number 1-10'''
--  return some_string[::-1] + str(np.random.randint(1,10))

-- Get the stage for your clean room.
CALL samooha_by_snowflake_local_db.provider.get_stage_for_python_files($cleanroom_name);

-- Save the file to the stage. Here is how to do it by using the Snowflake CLI
PUT file://~/reverser.py <STAGE_NAME> overwrite=True auto_compress=False;

-- Load the code from the stage into the clean room.
CALL samooha_by_snowflake_local_db.provider.load_python_into_cleanroom(
    $cleanroom_name,
    'reverse', -- Name used to call the function
    ['some_string  STRING'], -- Arguments and SQL types
    ['numpy'],               -- Any required packages
    ['/reverser.py'],        -- Relative path to file on stage
    'STRING',                -- Return type
    'reverser.main'          -- <FILE_NAME>.<FUNCTION_NAME>
);

-- Uploading code, even from a stage, increases the patch number.
CALL samooha_by_snowflake_local_db.provider.set_default_release_directive(
  $cleanroom_name, 'V1_0', <NEW_PATCH_NUMBER>);

-- Upload a template that calls the function.
CALL samooha_by_snowflake_local_db.provider.add_custom_sql_template(
    $cleanroom_name,
    $udf_template_name,
    $$
    SELECT
      p.status,
      cleanroom.reverse(p.status)
    FROM SAMOOHA_SAMPLE_DATABASE.DEMO.CUSTOMERS AS p
    LIMIT 100;
    $$
);

-- Switch to the consumer account and run the template to see the results.

The provider examples demonstrate uploading code from a stage.

Troubleshooting syntax errors or scan failures in uploaded code¶

If you upload a function that fails because of a syntax error, or if a security scan fails, an unpublishable patch can be generated. Therefore, you should thoroughly test your code before upload to ensure that it has no syntax errors.

You can see the list of packages, and their review status, by running the following SQL command, providing the clean room ID in the place indicated:

SHOW VERSIONS IN APPLICATION PACKAGE samooha_cleanroom_cleanroom_id;

Security scans¶

A security scan is run after any action that generates a new patch version in an external clean room, such as when the provider uploads Python into the clean room. (Consumer-submitted code, described on this page, does not trigger a security scan.) Internal clean rooms do not run security scans, but if you change an internal clean room to an external clean room, it will trigger a security scan for that patch. A clean roomm patch cannot be published externally until the patch has been scanned.

Snowflake Data Clean Rooms uses the Snowflake Native App security scan framework. Follow the native app security best practices to avoid security scan errors.

You can perform additional patch-creating actions before the last security scan is complete. However, you must wait for provider.view_cleanroom_scan_status to show success before you can update the default release directive in order to serve the latest version of the clean room.

Uploading multiple Python functions in a single patch¶

If you want to upload multiple template-callable Python packages to your clean room, you can call prepare_python_for_cleanroom multiple times, then call load_prepared_python_into_cleanroom once to scan, upload, and generate a single patch for your clean room. The following example demonstrates uploading a UDF and a UDTF using bulk uploading:

---- Add custom inline UDF ----
CALL samooha_by_snowflake_local_db.provider.prepare_python_for_cleanroom(
    $cleanroom_name,
    'get_next_status',  -- Name of the UDF. Can be different from the handler.
    ['status VARCHAR'], -- Arguments of the UDF, specified as (variable name, SQL type).
    ['numpy'],          -- Packages needed by UDF.
    [],                 -- When providing the code inline, this is an empty array.
    'VARCHAR',          -- Return type of UDF.
    'get_next_status',  -- Handler.
    $$
import numpy as np
def get_next_status(status):
  """Return the next higher status, or a random status
  if no matching status found or at the top of the list."""

  statuses = ['MEMBER', 'SILVER', 'GOLD', 'PLATINUM', 'DIAMOND']
  try:
    return statuses[statuses.index(status.upper()) + 1]
  except:
    return 'NO MATCH'
    $$
);

---- Add custom inline UDTF. ----
CALL samooha_by_snowflake_local_db.provider.prepare_python_for_cleanroom(
    $cleanroom_name,
    'get_info',  -- Name of the UDTF. Can be different from the handler.
    ['hashed_email VARCHAR', 'days_active INT', 'status VARCHAR', 'income VARCHAR'],   -- Name/Type arguments of the UDTF.
    ['numpy'],         -- Packages used by UDTF.
    [],                -- When providing the code inline, this is an empty array.
    'TABLE(hashed_email VARCHAR, months_active INT, level VARCHAR)',  -- Return type of UDTF.
    'GetSomeVals',     -- Handler class name.
$$
class GetSomeVals:
  def __init__(self):
    self.month_days = 30

  def process(self, hashed_email, days_active, status, income):
    '''Change days into rough months, and also return whether we
    think the user's membership status is lower, higher, or equal to
    what is expected, based on their income.'''

    months_active = days_active // self.month_days
    brackets = ['0-50K', '50K-100K', '100K-250K', '250K+']
    statuses = ['MEMBER', 'SILVER', 'GOLD', 'PLATINUM']
    if(statuses.index(status) < brackets.index(income)):
      level = 'low'
    elif(statuses.index(status) > brackets.index(income)):
      level = 'high'
    else:
      level = 'equal'

    yield(hashed_email, months_active, level)
$$
);

-- Upload all stored procedures.
-- Note the new patch number returned by this procedure. Keep this number for later use.
CALL samooha_by_snowflake_local_db.provider.load_prepared_python_into_cleanroom($cleanroom_name);

-- Set the release directive specified by the last load_python_into_cleanroom call.
CALL samooha_by_snowflake_local_db.provider.set_default_release_directive($cleanroom_name, 'V1_0', <PATCH_NUMBER>);

Provider-written code examples¶

The following examples demonstrate adding provider-written UDFs and UDTFs to a clean room.

Download the following examples and then upload them as worksheet files in your Snowflake account. You need separate accounts for the provider and consumer, each with the clean rooms API installed. Replace the information as noted in the sample files.

• Provider example code

• Consumer example code

• Loading a file from a stage. Run this notebook after you run the provider example to try loading a UDF from a stage.

• Uploading multiple Python functions in a single patch. This is a single-account internal testing clean room; you can use the same account for both the provider role and the consumer role.

Consumer-written code examples¶

The following examples demonstrate adding provider-written UDFs to a clean room.

Download the following examples and then upload them as worksheet files in your Snowflake account. You need separate accounts for the provider and consumer, each with the clean rooms API installed. Replace the information as noted in the sample files:

• Provider example code

• Consumer example code

• Consumer-written, provider-run code: Provider notebook

• Consumer-written, provider-run code: Consumer notebook