sdks/python/apache_beam/transforms/userstate.py

#
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements.  See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License.  You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#

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

Experimental; no backwards-compatibility guarantees.
"""

from __future__ import absolute_import

import types
from builtins import object

from apache_beam.coders import Coder
from apache_beam.coders import coders
from apache_beam.portability.api import beam_runner_api_pb2
from apache_beam.transforms.timeutil import TimeDomain


class StateSpec(object):
  """Specification for a user DoFn state cell."""

  def __init__(self):
    raise NotImplementedError

  def __repr__(self):
    return '%s(%s)' % (self.__class__.__name__, self.name)

  def to_runner_api(self, context):
    raise NotImplementedError


class BagStateSpec(StateSpec):
  """Specification for a user DoFn bag state cell."""

  def __init__(self, name, coder):
    assert isinstance(name, str)
    assert isinstance(coder, Coder)
    self.name = name
    self.coder = coder

  def to_runner_api(self, context):
    return beam_runner_api_pb2.StateSpec(
        bag_spec=beam_runner_api_pb2.BagStateSpec(
            element_coder_id=context.coders.get_id(self.coder)))


class SetStateSpec(StateSpec):
  """Specification for a user DoFn Set State cell"""

  def __init__(self, name, coder):
    if not isinstance(name, str):
      raise TypeError("SetState name is not a string")
    if not isinstance(coder, Coder):
      raise TypeError("SetState coder is not of type Coder")
    self.name = name
    self.coder = coder

  def to_runner_api(self, context):
    return beam_runner_api_pb2.StateSpec(
        set_spec=beam_runner_api_pb2.SetStateSpec(
            element_coder_id=context.coders.get_id(self.coder)))


class CombiningValueStateSpec(StateSpec):
  """Specification for a user DoFn combining value state cell."""

  def __init__(self, name, coder=None, combine_fn=None):
    """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.

    Args:
      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.
    """
    # Avoid circular import.
    from apache_beam.transforms.core import CombineFn
    # We want the coder to be optional, but unfortunately it comes
    # before the non-optional combine_fn parameter, which we can't
    # change for backwards compatibility reasons.
    #
    # Instead, allow it to be omitted (by either passing two arguments
    # or combine_fn by keyword.)
    if combine_fn is None:
      if coder is None:
        raise ValueError('combine_fn must be provided')
      else:
        coder, combine_fn = None, coder
    self.combine_fn = CombineFn.maybe_from_callable(combine_fn)
    if coder is None:
      coder = self.combine_fn.get_accumulator_coder()

    assert isinstance(name, str)
    assert isinstance(coder, Coder)
    self.name = name
    # The coder here should be for the accumulator type of the given CombineFn.
    self.coder = coder

  def to_runner_api(self, context):
    return beam_runner_api_pb2.StateSpec(
        combining_spec=beam_runner_api_pb2.CombiningStateSpec(
            combine_fn=self.combine_fn.to_runner_api(context),
            accumulator_coder_id=context.coders.get_id(self.coder)))


class TimerSpec(object):
  """Specification for a user stateful DoFn timer."""

  def __init__(self, name, time_domain):
    self.name = name
    if time_domain not in (TimeDomain.WATERMARK, TimeDomain.REAL_TIME):
      raise ValueError('Unsupported TimeDomain: %r.' % (time_domain,))
    self.time_domain = time_domain
    self._attached_callback = None

  def __repr__(self):
    return '%s(%s)' % (self.__class__.__name__, self.name)

  def to_runner_api(self, context):
    return beam_runner_api_pb2.TimerSpec(
        time_domain=TimeDomain.to_runner_api(self.time_domain),
        timer_coder_id=context.coders.get_id(
            coders._TimerCoder(coders.SingletonCoder(None))))


def on_timer(timer_spec):
  """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!')
  """

  if not isinstance(timer_spec, TimerSpec):
    raise ValueError('@on_timer decorator expected TimerSpec.')

  def _inner(method):
    if not callable(method):
      raise ValueError('@on_timer decorator expected callable.')
    if timer_spec._attached_callback:
      raise ValueError(
          'Multiple on_timer callbacks registered for %r.' % timer_spec)
    timer_spec._attached_callback = method
    return method

  return _inner


def get_dofn_specs(dofn):
  """Gets the state and timer specs for a DoFn, if any.

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

  # Avoid circular import.
  from apache_beam.runners.common import MethodWrapper
  from apache_beam.transforms.core import _DoFnParam
  from apache_beam.transforms.core import _StateDoFnParam
  from apache_beam.transforms.core import _TimerDoFnParam

  all_state_specs = set()
  all_timer_specs = set()

  # Validate params to process(), start_bundle(), finish_bundle() and to
  # any on_timer callbacks.
  for method_name in dir(dofn):
    if not isinstance(getattr(dofn, method_name, None), types.MethodType):
      continue
    method = MethodWrapper(dofn, method_name)
    param_ids = [d.param_id for d in method.defaults
                 if isinstance(d, _DoFnParam)]
    if len(param_ids) != len(set(param_ids)):
      raise ValueError(
          'DoFn %r has duplicate %s method parameters: %s.' % (
              dofn, method_name, param_ids))
    for d in method.defaults:
      if isinstance(d, _StateDoFnParam):
        all_state_specs.add(d.state_spec)
      elif isinstance(d, _TimerDoFnParam):
        all_timer_specs.add(d.timer_spec)

  return all_state_specs, all_timer_specs


def is_stateful_dofn(dofn):
  """Determines whether a given DoFn is a stateful DoFn."""

  # A Stateful DoFn is a DoFn that uses user state or timers.
  all_state_specs, all_timer_specs = get_dofn_specs(dofn)
  return bool(all_state_specs or all_timer_specs)


def validate_stateful_dofn(dofn):
  """Validates the proper specification of a stateful DoFn."""

  # Get state and timer specs.
  all_state_specs, all_timer_specs = get_dofn_specs(dofn)

  # Reject DoFns that have multiple state or timer specs with the same name.
  if len(all_state_specs) != len(set(s.name for s in all_state_specs)):
    raise ValueError(
        'DoFn %r has multiple StateSpecs with the same name: %s.' % (
            dofn, all_state_specs))
  if len(all_timer_specs) != len(set(s.name for s in all_timer_specs)):
    raise ValueError(
        'DoFn %r has multiple TimerSpecs with the same name: %s.' % (
            dofn, all_timer_specs))

  # Reject DoFns that use timer specs without corresponding timer callbacks.
  for timer_spec in all_timer_specs:
    if not timer_spec._attached_callback:
      raise ValueError(
          ('DoFn %r has a TimerSpec without an associated on_timer '
           'callback: %s.') % (dofn, timer_spec))
    method_name = timer_spec._attached_callback.__name__
    if (timer_spec._attached_callback !=
        getattr(dofn, method_name, None).__func__):
      raise ValueError(
          ('The on_timer callback for %s is not the specified .%s method '
           'for DoFn %r (perhaps it was overwritten?).') % (
               timer_spec, method_name, dofn))


class RuntimeTimer(object):
  """Timer interface object passed to user code."""

  def __init__(self, timer_spec):
    self._cleared = False
    self._new_timestamp = None

  def clear(self):
    self._cleared = True
    self._new_timestamp = None

  def set(self, timestamp):
    self._new_timestamp = timestamp


class RuntimeState(object):
  """State interface object passed to user code."""
  def prefetch(self):
    # The default implementation here does nothing.
    pass


class AccumulatingRuntimeState(RuntimeState):
  def read(self):
    raise NotImplementedError(type(self))

  def add(self, value):
    raise NotImplementedError(type(self))

  def clear(self):
    raise NotImplementedError(type(self))


class BagRuntimeState(AccumulatingRuntimeState):
  """Bag state interface object passed to user code."""


class SetRuntimeState(AccumulatingRuntimeState):
  """Set state interface object passed to user code."""


class CombiningValueRuntimeState(AccumulatingRuntimeState):
  """Combining value state interface object passed to user code."""


class UserStateContext(object):
  """Wrapper allowing user state and timers to be accessed by a DoFnInvoker."""

  def get_timer(self, timer_spec, key, window):
    raise NotImplementedError(type(self))

  def get_state(self, state_spec, key, window):
    raise NotImplementedError(type(self))

  def commit(self):
    raise NotImplementedError(type(self))