UDF API List

This section describes the external APIs provided by the UDF module. You can call these APIs to develop custom processing functions, and then execute such functions on the CPU through DataFlow graph construction.

You can view the API implementation in python/site-packages/dataflow/flow_func/flow_func.py in CANN component directory. The APIs are as follows.

Class FlowMsg

This class is used to process the input and output operations of FlowFunc.

**Table 1** FlowMsg APIs
API	Description
FlowMsg Constructor	Constructs an object of class FlowMsg.
get_msg_type	Obtains the message type of FlowMsg.
get_tensor	Obtains the tensor object of FlowMsg.
set_ret_code	Sets the error code in the FlowMsg message.
get_ret_code	Obtains the error code in the input FlowMsg message.
set_start_time	Sets the start timestamp in the FlowMsg message header.
get_start_time	Obtains the start timestamp in the FlowMsg message.
set_end_time	Sets the end timestamp in the FlowMsg message header.
get_end_time	Obtains the end timestamp in the FlowMsg message.
set_flow_flags	Sets flags in the FlowMsg message header.
get_flow_flags	Obtains flags in the FlowMsg message header.
set_route_label	Sets route labels.
get_transaction_id	Obtains the transaction ID in the FlowMsg message. The transaction ID starts from 1, and is incremented by 1 each time a batch of data is fed. This function can be used to identify the ID of the data batch.
set_msg_type	Sets the message type of FlowMsg.
get_raw_data	Obtains data of the rawdata type.
set_transaction_id	Sets the transaction ID used for DataFlow data transmission.

Tensor Class

This class is used to perform tensor-related operations. The obtained tensor is dataflow.Tensor.

**Table 2** Tensor APIs
API	Description
Tensor Constructor	Constructs an object of class Tensor.
get_shape	Obtains the shape of a tensor.
get_data_type	Obtains the data type of a tensor.
get_data_size	Obtains the data size of a tensor.
get_element_cnt	Obtains the number of elements in a tensor.
reshape	Reshapes a tensor without changing the tensor content.

Class MetaParams

This class is used to obtain shared variable information.

**Table 3** MetaParams APIs
API	Description
PyMetaParams Constructor	Constructs an object of class PyMetaParams.
get_name	Obtains the instance name of FlowFunc.
get_attr_int	Obtains the attribute value of the int type with the specified name.
get_attr_bool_list	Obtains the attribute value of the bool array with the specified name.
get_attr_int_list	Obtains the attribute value of the int array with the specified name.
get_attr_int_list_list	Obtains the attribute value of the 2D array of the int type with the specified name.
get_attr_bool	Obtains the attribute value of the bool type with the specified name.
get_attr_float_list	Obtains the attribute value of the float array with the specified name.
get_attr_tensor_dtype	Obtains the attribute value of the numpy.dtype type with the specified name.
get_attr_tensor_dtype_list	Obtains the attribute value of the array of the numpy.dtype types with the specified name.
get_attr_str	Obtains the attribute value of the string type with the specified name.
get_attr_str_list	Obtains the attribute value of the string array with the specified name.
get_attr_float	Obtains the attribute value of the float type with the specified name.
get_input_num	Obtains the number of FlowFunc inputs.
get_output_num	Obtains the number of FlowFunc outputs.
get_work_path	Obtains the working path of FlowFunc.
get_running_device_id	Obtains the ID of the running device.

Class MetaRunContext

This class is used to process the context information of the processing function of FlowFunc, such as allocating a tensor, setting the output, and running FlowModel.

**Table 4** MetaRunContext APIs
API	Description
MetaRunContext Constructor	Constructs an object of class MetaRunContext.
alloc_tensor_msg	Allocates FlowMsg of the tensor type based on the shape, data type, and alignment size.
set_output	Sets the tensor of the output with a specified index.
set_multi_outputs	Sets the tensors of the outputs with specified indexes in batches.
run_flow_model	Runs a specified model synchronously.
alloc_empty_data_msg	Allocates a message of the MsgType type that carries empty data.
get_user_data	Obtains the user-defined data.
raise_exception	The UDF proactively reports an exception.
get_exception	The UDF obtains the exception reported by other UDF nodes.
alloc_raw_data_msg	Allocates a continuous memory based on the input size to carry the FlowMsg of the raw data type.
to_flow_msg	Converts a dataflow tensor into a FlowMsg.

Class AffinityPolicy

**Table 5** AffinityPolicy APIs
API	Description
Class AffinityPolicy	Defines the affinity policy enumeration.

Class BalanceConfig

When balanced distribution is required, you need to configure the output data identifier and weight matrix. The scheduling module can implement balanced distribution among multiple instances based on the configuration.

**Table 6** BalanceConfig APIs
API	Description
BalanceConfig Constructor	Constructs an object of class BalanceConfig.
set_data_pos	Sets the position of the output data in the weight matrix.
get_inner_config	Obtains inner configuration objects. This API is called by set_output or set_multi_outputs.

Class FlowMsgQueue

This class serves as the input queue for FlowFunc in streaming input scenarios (the input parameter of FlowFunc is a queue). It dequeues FlowMsg entries, converts each to its corresponding data type based on MsgType, and returns the converted value to the caller.

**Table 7** FlowMsgQueue APIs
API	Description
FlowMsgQueue Constructor	Constructs and destructs an object of class FlowMsgQueue.
get	Obtains elements in a queue.
get_nowait	Obtains elements in a queue without waiting. The function is the same as get(block=False).
full	Checks whether the queue is full.
full	Checks whether the queue is empty.
qsize	Obtains the number of current elements in a queue.

UDF Log APIs

Logging APIs are exposed for Python UDFs. You need to import the flow_func module to use these APIs. Use the defined logger object within this module to call its encapsulated logging APIs of different severity levels.

**Table 8** UDF log APIs
API	Description
FlowFuncLogger Constructor	Constructs an object of class FlowFuncLogger.
get_log_header	Obtains log extension header information.
is_log_enable	Checks whether logs of the corresponding level and type are enabled.
Error-Level Log Macro of Run Logs	Defines the error-level log macro of run logs.
Info-Level Log Macro of Run Logs	Defines the info-level log macro of run logs.
Error-Level Log Macro of Debug Logs	Defines the error-level log macro of debug logs.
Warn-Level Log Macro of Debug Logs	Defines the warn-level log macro of debug logs.
Info-Level Log Macro of Debug Logs	Defines the info-level log macro of debug logs.
Debug-Level Log Macro of Debug Logs	Defines the debug-level log macro of debug logs.

Parent topic: UDF APIs