Create a Snowpark project definition¶

The snowflake.yml file contains the functions and procedures declarations for a Snowpark project.

Note

Currently, the Snowpark project definition file must be named snowflake.yml.

The following snippet shows a sample Snowpark project definition file: with two functions and two procedures. The hello_function function uses external capabilities of Snowpark.

definition_version: 1
snowpark:
  project_name: "my_snowpark_project"
  stage_name: "dev_deployment"
  src: "app/"

  functions:
    - name: hello_function
      handler: "functions.hello_function"
      signature:
        - name: "name"
          type: "string"
      returns: string
      external_access_integration:
        - my_external_access
      secrets:
        cred: my_cred_name
    - name: test_function
      handler: "functions.hello_function"
      signature: ""
      returns: string

  procedures:
    - name: hello_procedure
      handler: "procedures.hello_procedure"
      signature:
        - name: "name"
          type: "string"
      returns: string
    - name: test_procedure
      handler: "procedures.test_procedure"
      signature: ""
      returns: string

Copy

Caution

Files inside a project directory are processed by Snowflake CLI and could be uploaded to Snowflake when executing other snow snowpark commands. You should use caution when putting any sensitive information inside files in a project directory.

Project definition properties¶

The following table describes the project definition properties.

Project definition properties¶
Property	Definition
definition_version required, int	Version of the project definition schema, which is currently 1.
snowpark.project_name required, string	Project identifier.
snowpark.stage_name required, string	Stage in which project’s artifacts will be stored.
snowpark.src required, int	The `src` parameter specifies a folder where your code should be located. You can think about it as a Python module. This path is used by the `snow snowpark build` command to build the zip package (which name matches the folder name) as well as by `snow snowpark deploy` to deploy the artifact.
snowpark.functions optional, int	List of functions defined in the project. See Function and procedure object properties for more details.
snowpark.procedures optional, int	List of procedures defined in the project. See Function and procedure object properties for more details.

Function and procedure object properties¶

The following table describes the properties used by functions and procedures.

Function and procedure object properties¶
Property	Definition
name required, string	Object identifier. You can specify either the name or the fully-qualified name of the function or procedure. If you specify a fully-qualified name, such as `mydb.schema1.my-function`, you cannot specify values for the database and schema properties in the project definition. If you supply both, an error occurs. functions: # valid fully-qualified function name # with no database or schema properties - name: "mydb.schema1.my-function" # valid function name with # database and schema properties - name: "my-function" database: "mydb" schema: "schema1" # invalid fully-qualified function name # with database and schema properties - name: "mydb.schema1.my-function" database: "mydb" schema: "schema1" Copy
handler required, string	Function’s or procedure’s implementation of the object inside module defined in `snowpark.src`. For example `functions.hello_function` refers to function `hello_function` from file `<src>/functions.py`.
returns required, string	Type of the result. Check the list of available types.
signature required, sequence	The `signature` parameter describes consecutive arguments passed to the object. Each should specify its name and type, for example: signature: - name: "first_argument" type: int - name: "second_argument" default: "default value" type: string Copy If a function or procedure takes no arguments, set this value to an empty string (`signature: ""`). Check the list of available types. To learn more about the syntax of named and optional arguments, see Calling a UDF That Has Optional Arguments.
runtime optional, string	Python version to use when executing the procedure or function. Default: “3.8”.
external_access_integration optional, string sequence	Names of external access integrations needed for this procedure’s handler code to access external networks. See CREATE PROCEDURE for more details.
database optional, string	Name of the database for the function or procedure. Default: None. An error occurs if you specify this property and use a fully-qualified name in the `name` property (such as `mydb.schema1.my-function`).
schema optional, string	Name of the schema for the function or procedure. Default: None. An error occurs if you specify this property and use a fully-qualified name in the `name` property (such as `mydb.schema1.my-function`).
secrets optional, dictionary	Assigns the names of secrets to variables so that you can use the variables to reference the secrets when retrieving information from secrets in handler code. See CREATE PROCEDURE for more details.
imports optional, string sequence	Stage and path to previously uploaded files you want to import. See CREATE PROCEDURE for more details.
execute_as_caller optional, bool	Available only for procedures. Determine whether the procedure is executed with the privileges of the owner (you) or with the privileges of the caller. Default: False (owner’s privileges).