UserDataFunctions Class

This class is necessary to define User Data Functions in Fabric. Please ensure an instantiation of this class exists in your code before any User Data Functions are defined.

Constructor

UserDataFunctions()

Remarks

This class is used to define User Data Functions in Fabric. The class must be instantiated before any User Data Functions are defined. The instantiation of this class is required.

Methods

alias_connection	This decorator is used to tell a User Data Function that there is a connection to a data source. This decorator must be used in tandem with a parameter of type FabricSqlConnection or FabricLakehouseClient (see the example under Remarks).
connection	This decorator is used to tell a User Data Function that there is a connection to a data source. This decorator must be used in tandem with a parameter of type FabricSqlConnection or FabricLakehouseClient (see the example under Remarks).
context	This decorator is used to tell a User Data Function that there is a UserDataFunctionContext parameter. This decorator must be used in tandem with a parameter of type UserDataFunctionContext.
fabric_item_input	This decorator is deprecated. Please use connection instead. Note the alias parameter in connection is now the first parameter.
function	This decorator is used to define a User Data Function in Fabric. The function must be decorated with this decorator in order to be recognized as a User Data Function.
generic_connection	This decorator is used to tell a User Data Function that there is a connection to a data source. This decorator must be used in tandem with a parameter of type FabricSqlConnection or FabricLakehouseClient (see the example under Remarks).
get_udf_oai_spec
register_needed_modules	Register needed modules for the specified audience type. If the audience type is 'CosmosDB', the CosmosDB module will be imported. If the audience type is 'KeyVault', the KeyVault module will be imported. More will be added as needed.
register_provider
streaming_function	Define a streaming User Data Function whose HTTP response is flushed to the caller incrementally instead of being buffered into a single body. The decorated function must return a StreamResponse wrapping an iterator (sync or async) of `bytes`/`str` chunks. Each chunk is streamed to the client as it is produced — enabling true end-to-end HTTP streaming (e.g. relaying an Apache Arrow stream from Power BI `executeDaxQueries`). Important Streaming uses the Azure Functions HTTP streams feature and requires the azurefunctions-extensions-http-fastapi package in your requirements.txt plus the app setting PYTHON_ENABLE_INIT_INDEXING=1. When a streaming function is present the app runs in ASGI mode app-wide, so it cannot also host classic (buffered) @udf.function() functions in the same app.
user_data_function_context_input	This decorator is deprecated. Please use context instead.

alias_connection

This decorator is used to tell a User Data Function that there is a connection to a data source. This decorator must be used in tandem with a parameter of type FabricSqlConnection or FabricLakehouseClient (see the example under Remarks).

alias_connection(alias: str, argName: str | None = None, **kwargs) -> Callable[[...], Any]

Parameters

Name	Description
alias Required	str The alias of the data connection that is being used.
argName	str The name of the parameter in the function signature. If not provided, the alias will be used. Default value: None

connection

connection(alias: str | None = None, argName: str | None = None, **kwargs) -> Callable[[...], Any]

Parameters

Name	Description
alias	str The alias of the data connection that is being used. Default value: None
argName	str The name of the parameter in the function signature. If not provided, the alias will be used. Default value: None

context

This decorator is used to tell a User Data Function that there is a UserDataFunctionContext parameter. This decorator must be used in tandem with a parameter of type UserDataFunctionContext.

context(argName, **kwargs) -> Callable[[...], Any]

Parameters

Name	Description
argName Required	str The name of the parameter in the function signature.

fabric_item_input

This decorator is deprecated. Please use connection instead. Note the alias parameter in connection is now the first parameter.

fabric_item_input(argName, alias: str, **kwargs) -> Callable[[...], Any]

Parameters

Name	Description
argName Required
alias Required

function

This decorator is used to define a User Data Function in Fabric. The function must be decorated with this decorator in order to be recognized as a User Data Function.

function(name=None)

Parameters

Name	Description
name	str The name of the function. This parameter is not used in the current version of Fabric. Default value: None

generic_connection

generic_connection(argName: str, **kwargs) -> Callable[[...], Any]

Parameters

Name	Description
audienceType Required	str The audience type of the data connection that is being used.
argName Required	str The name of the parameter in the function signature. If not provided, the audience type will be used.

get_udf_oai_spec

get_udf_oai_spec(publicUrl: str, privateFunctions: list, specFormat: str = 'yaml') -> str

Parameters

Name	Description
publicUrl Required
privateFunctions Required
specFormat	Default value: yaml

register_needed_modules

If the audience type is 'CosmosDB', the CosmosDB module will be imported. If the audience type is 'KeyVault', the KeyVault module will be imported.

More will be added as needed.

register_needed_modules(audienceType: str)

Parameters

Name	Description
audienceType Required

register_provider

register_provider(type: Type, provider: Type)

Parameters

Name	Description
type Required
provider Required

streaming_function

Define a streaming User Data Function whose HTTP response is flushed to the caller incrementally instead of being buffered into a single body.

The decorated function must return a StreamResponse wrapping an iterator (sync or async) of bytes/str chunks. Each chunk is streamed to the client as it is produced — enabling true end-to-end HTTP streaming (e.g. relaying an Apache Arrow stream from Power BI executeDaxQueries).

Important

Streaming uses the Azure Functions HTTP streams feature and

requires the azurefunctions-extensions-http-fastapi package in

your requirements.txt plus the app setting

PYTHON_ENABLE_INIT_INDEXING=1. When a streaming function is

present the app runs in ASGI mode app-wide, so it cannot also host

classic (buffered) @udf.function() functions in the same app.

streaming_function(name=None)

Parameters

Name	Description
name	str Optional function/route name. Defaults to the function name. Default value: None

user_data_function_context_input

This decorator is deprecated. Please use context instead.

user_data_function_context_input(argName, **kwargs) -> Callable[[...], Any]

Parameters

Name	Description
argName Required