apache_beam.transforms.userstate module

User-facing interfaces for the Beam State and Timer APIs.

class apache_beam.transforms.userstate.StateSpec(name: str, coder: Coder)

Bases: object

Specification for a user DoFn state cell.

to_runner_api(context: PipelineContext) → StateSpec

class apache_beam.transforms.userstate.ReadModifyWriteStateSpec(name: str, coder: Coder)

Bases: StateSpec

Specification for a user DoFn value state cell.

to_runner_api(context: PipelineContext) → StateSpec

class apache_beam.transforms.userstate.BagStateSpec(name: str, coder: Coder)

Bases: StateSpec

Specification for a user DoFn bag state cell.

to_runner_api(context: PipelineContext) → StateSpec

class apache_beam.transforms.userstate.SetStateSpec(name: str, coder: Coder)

Bases: StateSpec

Specification for a user DoFn Set State cell

to_runner_api(context: PipelineContext) → StateSpec

class apache_beam.transforms.userstate.CombiningValueStateSpec(name: str, coder: Coder | None = None, combine_fn: Any | None = None)

Bases: StateSpec

Specification for a user DoFn combining value state cell.

Initialize the specification for CombiningValue state.

CombiningValueStateSpec(name, combine_fn) -> Coder-inferred combining value

state spec.

CombiningValueStateSpec(name, coder, combine_fn) -> Combining value state

spec with coder and combine_fn specified.

Parameters:

• name (str) – The name by which the state is identified.

• coder (Coder) – Coder specifying how to encode the values to be combined. May be inferred.

• combine_fn (CombineFn or callable) – Function specifying how to combine the values passed to state.

to_runner_api(context: PipelineContext) → StateSpec

class apache_beam.transforms.userstate.OrderedListStateSpec(name: str, coder: Coder)

Bases: StateSpec

Specification for a user DoFn ordered list state cell.

to_runner_api(context: PipelineContext) → StateSpec

class apache_beam.transforms.userstate.Timer(user_key, dynamic_timer_tag, windows, clear_bit, fire_timestamp, hold_timestamp, paneinfo)

Bases: tuple

Create new instance of Timer(user_key, dynamic_timer_tag, windows, clear_bit, fire_timestamp, hold_timestamp, paneinfo)

clear_bit: bool

Alias for field number 3

dynamic_timer_tag: str

Alias for field number 1

fire_timestamp: Timestamp | None

Alias for field number 4

hold_timestamp: Timestamp | None

Alias for field number 5

paneinfo: windowed_value.PaneInfo | None

Alias for field number 6

user_key: Any

Alias for field number 0

windows: Tuple[windowed_value.BoundedWindow, ...]

Alias for field number 2

class apache_beam.transforms.userstate.TimerSpec(name: str, time_domain: str)

Bases: object

Specification for a user stateful DoFn timer.

prefix = 'ts-'

to_runner_api(context: PipelineContext, key_coder: Coder, window_coder: Coder) → TimerFamilySpec

apache_beam.transforms.userstate.on_timer(timer_spec: TimerSpec) → Callable[[CallableT], CallableT]

Decorator for timer firing DoFn method.

This decorator allows a user to specify an on_timer processing method in a stateful DoFn. Sample usage:

class MyDoFn(DoFn):
  TIMER_SPEC = TimerSpec('timer', TimeDomain.WATERMARK)

  @on_timer(TIMER_SPEC)
  def my_timer_expiry_callback(self):
    logging.info('Timer expired!')

apache_beam.transforms.userstate.get_dofn_specs(dofn: DoFn) → Tuple[Set[StateSpec], Set[TimerSpec]]

Gets the state and timer specs for a DoFn, if any.

Parameters:

dofn (apache_beam.transforms.core.DoFn) – The DoFn instance to introspect for timer and state specs.

apache_beam.transforms.userstate.is_stateful_dofn(dofn: DoFn) → bool

Determines whether a given DoFn is a stateful DoFn.

apache_beam.transforms.userstate.validate_stateful_dofn(dofn: DoFn) → None

Validates the proper specification of a stateful DoFn.

class apache_beam.transforms.userstate.BaseTimer

Bases: object

clear(dynamic_timer_tag: str = '') → None

set(timestamp: Timestamp, dynamic_timer_tag: str = '') → None

class apache_beam.transforms.userstate.RuntimeTimer

Bases: BaseTimer

Timer interface object passed to user code.

clear(dynamic_timer_tag: str = '') → None

set(timestamp: Timestamp, dynamic_timer_tag: str = '') → None

class apache_beam.transforms.userstate.RuntimeState

Bases: object

State interface object passed to user code.

prefetch() → None

finalize() → None

class apache_beam.transforms.userstate.ReadModifyWriteRuntimeState

Bases: RuntimeState

read() → Any

write(value: Any) → None

clear() → None

commit() → None

class apache_beam.transforms.userstate.AccumulatingRuntimeState

Bases: RuntimeState

read() → Iterable[Any]

add(value: Any) → None

clear() → None

commit() → None

class apache_beam.transforms.userstate.BagRuntimeState

Bases: AccumulatingRuntimeState

Bag state interface object passed to user code.

class apache_beam.transforms.userstate.SetRuntimeState

Bases: AccumulatingRuntimeState

Set state interface object passed to user code.

class apache_beam.transforms.userstate.CombiningValueRuntimeState

Bases: AccumulatingRuntimeState

Combining value state interface object passed to user code.

class apache_beam.transforms.userstate.OrderedListRuntimeState

Bases: AccumulatingRuntimeState

Ordered list state interface object passed to user code.

read() → Iterable[Tuple[Timestamp, Any]]

add(value: Tuple[Timestamp, Any]) → None

read_range(min_time_stamp: Timestamp, limit_time_stamp: Timestamp) → Iterable[Tuple[Timestamp, Any]]

clear_range(min_time_stamp: Timestamp, limit_time_stamp: Timestamp) → None

class apache_beam.transforms.userstate.UserStateContext

Bases: object

Wrapper allowing user state and timers to be accessed by a DoFnInvoker.

get_timer(timer_spec: TimerSpec, key: Any, window: windowed_value.BoundedWindow, timestamp: Timestamp, pane: PaneInfo) → BaseTimer

get_state(state_spec: StateSpec, key: Any, window: windowed_value.BoundedWindow) → RuntimeState

commit() → None