예제 #1
0
class Agent(ABC):
    """
    This class encompasses a worker as they are working on an individual assignment.
    It maintains details for the current task at hand such as start and end time,
    connection status, etc.
    """
    def __init__(self,
                 db: "MephistoDB",
                 db_id: str,
                 row: Optional[Mapping[str, Any]] = None):
        self.db: "MephistoDB" = db
        if row is None:
            row = db.get_agent(db_id)
        assert row is not None, f"Given db_id {db_id} did not exist in given db"
        self.db_id: str = row["agent_id"]
        self.db_status = row["status"]
        self.worker_id = row["worker_id"]
        self.unit_id = row["unit_id"]
        self.task_type = row["task_type"]
        self.provider_type = row["provider_type"]
        self.pending_observations: List["Packet"] = []
        self.pending_actions: List["Packet"] = []
        self.has_action = threading.Event()
        self.has_action.clear()
        self.wants_action = threading.Event()
        self.wants_action.clear()
        self.has_updated_status = threading.Event()
        self.assignment_id = row["assignment_id"]
        self.task_run_id = row["task_run_id"]
        self.task_id = row["task_id"]
        self.did_submit = threading.Event()

        # Deferred loading of related entities
        self._worker: Optional["Worker"] = None
        self._unit: Optional["Unit"] = None
        self._assignment: Optional["Assignment"] = None
        self._task_run: Optional["TaskRun"] = None
        self._task: Optional["Task"] = None

        # Follow-up initialization
        self.state = AgentState(self)  # type: ignore

    def __new__(cls,
                db: "MephistoDB",
                db_id: str,
                row: Optional[Mapping[str, Any]] = None) -> "Agent":
        """
        The new method is overridden to be able to automatically generate
        the expected Agent class without needing to specifically find it
        for a given db_id. As such it is impossible to create a base Agent
        as you will instead be returned the correct Agent class according to
        the crowdprovider associated with this Agent.
        """
        from mephisto.core.registry import get_crowd_provider_from_type

        if cls == Agent:
            # We are trying to construct a Agent, find what type to use and
            # create that instead
            if row is None:
                row = db.get_agent(db_id)
            assert row is not None, f"Given db_id {db_id} did not exist in given db"
            correct_class = get_crowd_provider_from_type(
                row["provider_type"]).AgentClass
            return super().__new__(correct_class)
        else:
            # We are constructing another instance directly
            return super().__new__(cls)

    def get_agent_id(self) -> str:
        """Return this agent's id"""
        return self.db_id

    def get_worker(self) -> Worker:
        """
        Return the worker that is using this agent for a task
        """
        if self._worker is None:
            self._worker = Worker(self.db, self.worker_id)
        return self._worker

    def get_unit(self) -> "Unit":
        """
        Return the Unit that this agent is working on.
        """
        if self._unit is None:
            from mephisto.data_model.assignment import Unit

            self._unit = Unit(self.db, self.unit_id)
        return self._unit

    def get_assignment(self) -> "Assignment":
        """Return the assignment this agent is working on"""
        if self._assignment is None:
            if self._unit is not None:
                self._assignment = self._unit.get_assignment()
            else:
                from mephisto.data_model.assignment import Assignment

                self._assignment = Assignment(self.db, self.assignment_id)
        return self._assignment

    def get_task_run(self) -> "TaskRun":
        """Return the TaskRun this agent is working within"""
        if self._task_run is None:
            if self._unit is not None:
                self._task_run = self._unit.get_task_run()
            elif self._assignment is not None:
                self._task_run = self._assignment.get_task_run()
            else:
                from mephisto.data_model.task import TaskRun

                self._task_run = TaskRun(self.db, self.task_run_id)
        return self._task_run

    def get_task(self) -> "Task":
        """Return the Task this agent is working within"""
        if self._task is None:
            if self._unit is not None:
                self._task = self._unit.get_task()
            elif self._assignment is not None:
                self._task = self._assignment.get_task()
            elif self._task_run is not None:
                self._task = self._task_run.get_task()
            else:
                from mephisto.data_model.task import Task

                self._task = Task(self.db, self.task_id)
        return self._task

    def get_data_dir(self) -> str:
        """
        Return the directory to be storing any agent state for
        this agent into
        """
        assignment_dir = self.get_assignment().get_data_dir()
        return os.path.join(assignment_dir, self.db_id)

    def update_status(self, new_status: str) -> None:
        """Update the database status of this agent, and
        possibly send a message to the frontend agent informing
        them of this update"""
        if self.db_status == new_status:
            return  # Noop, this is already the case
        if self.db_status in AgentState.complete():
            print(f"Updating a final status, was {self.db_status} "
                  f"and want to set to {new_status}")
        self.db.update_agent(self.db_id, status=new_status)
        self.db_status = new_status
        self.has_updated_status.set()
        if new_status in [
                AgentState.STATUS_RETURNED, AgentState.STATUS_DISCONNECT
        ]:
            # Disconnect statuses should free any pending acts
            self.has_action.set()
            self.did_submit.set()

    @staticmethod
    def _register_agent(db: "MephistoDB", worker: Worker, unit: "Unit",
                        provider_type: str) -> "Agent":
        """
        Create this agent in the mephisto db with the correct setup
        """
        db_id = db.new_agent(
            worker.db_id,
            unit.db_id,
            unit.task_id,
            unit.task_run_id,
            unit.assignment_id,
            unit.task_type,
            provider_type,
        )
        a = Agent(db, db_id)
        a.update_status(AgentState.STATUS_ACCEPTED)
        return a

    # Specialized child cases may need to implement the following

    @classmethod
    def new_from_provider_data(
        cls,
        db: "MephistoDB",
        worker: Worker,
        unit: "Unit",
        provider_data: Dict[str, Any],
    ) -> "Agent":
        """
        Wrapper around the new method that allows registering additional
        bookkeeping information from a crowd provider for this agent
        """
        agent = cls.new(db, worker, unit)
        unit.worker_id = worker.db_id
        agent._unit = unit
        return agent

    def observe(self, packet: "Packet") -> None:
        """
        Pass the observed information to the AgentState, then
        queue the information to be pushed to the user
        """
        sending_packet = packet.copy()
        sending_packet.receiver_id = self.db_id
        self.state.update_data(sending_packet)
        self.pending_observations.append(sending_packet)

    def act(self, timeout: Optional[int] = None) -> Optional["Packet"]:
        """
        Request information from the Agent's frontend. If non-blocking,
        (timeout is None) should return None if no actions are ready
        to be returned.
        """
        if len(self.pending_actions) == 0:
            self.wants_action.set()
            if timeout is None or timeout == 0:
                return None
            self.has_action.wait(timeout)

        if len(self.pending_actions) == 0:
            # various disconnect cases
            status = self.get_status()
            if status == AgentState.STATUS_DISCONNECT:
                raise AgentDisconnectedError(self.db_id)
            elif status == AgentState.STATUS_RETURNED:
                raise AgentReturnedError(self.db_id)
            self.update_status(AgentState.STATUS_TIMEOUT)
            raise AgentTimeoutError(timeout, self.db_id)
        assert len(
            self.pending_actions) > 0, "has_action released without an action!"

        act = self.pending_actions.pop(0)

        if "MEPHISTO_is_submit" in act.data and act.data["MEPHISTO_is_submit"]:
            self.did_submit.set()

        if len(self.pending_actions) == 0:
            self.has_action.clear()
        self.state.update_data(act)
        return act

    def get_status(self) -> str:
        """Get the status of this agent in their work on their unit"""
        if self.db_status not in AgentState.complete():
            row = self.db.get_agent(self.db_id)
            if row["status"] != self.db_status:
                if row["status"] in [
                        AgentState.STATUS_RETURNED,
                        AgentState.STATUS_DISCONNECT,
                ]:
                    # Disconnect statuses should free any pending acts
                    self.has_action.set()
                self.has_updated_status.set()
            self.db_status = row["status"]
        return self.db_status

    # Children classes should implement the following methods

    def approve_work(self) -> None:
        """Approve the work done on this agent's specific Unit"""
        raise NotImplementedError()

    def soft_reject_work(self) -> None:
        """
        Pay a worker for attempted work, but mark it as below the 
        quality bar for this assignment
        """
        # TODO(OWN) extend this method to assign a soft block
        # qualification automatically if a threshold of
        # soft rejects as a proportion of total accepts
        # is exceeded
        self.approve_work()
        self.update_status(AgentState.STATUS_SOFT_REJECTED)

    def reject_work(self, reason) -> None:
        """Reject the work done on this agent's specific Unit"""
        raise NotImplementedError()

    def mark_done(self) -> None:
        """
        Take any required step with the crowd_provider to ensure that
        the worker can submit their work and be marked as complete via
        a call to get_status
        """
        raise NotImplementedError()

    @staticmethod
    def new(db: "MephistoDB", worker: Worker, unit: "Unit") -> "Agent":
        """
        Create an agent for this worker to be used for work on the given Unit.

        Implementation should return the result of _register_agent when sure the agent
        can be successfully created to have it put into the db.
        """
        raise NotImplementedError()
예제 #2
0
class Unit(ABC):
    """
    This class tracks the status of an individual worker's contribution to a
    higher level assignment. It is the smallest 'unit' of work to complete
    the assignment, and this class is only responsible for checking
    the status of that work itself being done.

    It should be extended for usage with a specific crowd provider
    """
    def __init__(self,
                 db: "MephistoDB",
                 db_id: str,
                 row: Optional[Mapping[str, Any]] = None):
        self.db: "MephistoDB" = db
        if row is None:
            row = db.get_unit(db_id)
        assert row is not None, f"Given db_id {db_id} did not exist in given db"
        self.db_id: str = row["unit_id"]
        self.assignment_id = row["assignment_id"]
        self.unit_index = row["unit_index"]
        self.pay_amount = row["pay_amount"]
        self.agent_id = row["agent_id"]
        self.provider_type = row["provider_type"]
        self.db_status = row["status"]
        self.task_type = row["task_type"]
        self.task_id = row["task_id"]
        self.task_run_id = row["task_run_id"]
        self.sandbox = row["sandbox"]
        self.requester_id = row["requester_id"]
        self.worker_id = row["worker_id"]

        # Deferred loading of related entities
        self.__task: Optional["Task"] = None
        self.__task_run: Optional["TaskRun"] = None
        self.__assignment: Optional["Assignment"] = None
        self.__requester: Optional["Requester"] = None
        self.__agent: Optional["Agent"] = None
        self.__worker: Optional["Worker"] = None

    def __new__(cls,
                db: "MephistoDB",
                db_id: str,
                row: Optional[Mapping[str, Any]] = None) -> "Unit":
        """
        The new method is overridden to be able to automatically generate
        the expected Unit class without needing to specifically find it
        for a given db_id. As such it is impossible to create a Unit
        as you will instead be returned the correct Unit class according to
        the crowdprovider associated with this Unit.
        """
        if cls == Unit:
            # We are trying to construct a Unit, find what type to use and
            # create that instead
            from mephisto.operations.registry import get_crowd_provider_from_type

            if row is None:
                row = db.get_unit(db_id)
            assert row is not None, f"Given db_id {db_id} did not exist in given db"
            correct_class = get_crowd_provider_from_type(
                row["provider_type"]).UnitClass
            return super().__new__(correct_class)
        else:
            # We are constructing another instance directly
            return super().__new__(cls)

    def get_crowd_provider_class(self) -> Type["CrowdProvider"]:
        """Get the CrowdProvider class that manages this Unit"""
        from mephisto.operations.registry import get_crowd_provider_from_type

        return get_crowd_provider_from_type(self.provider_type)

    def get_assignment_data(self) -> Optional[Dict[str, Any]]:
        """Return the specific assignment data for this assignment"""
        return self.get_assignment().get_assignment_data()

    def sync_status(self) -> None:
        """
        Ensure that the queried status from this unit and the db status
        are up to date
        """
        # TODO(102) this will need to be run periodically/on crashes
        # to sync any lost state
        self.set_db_status(self.get_status())

    def get_db_status(self) -> str:
        """
        Return the status as currently stored in the database
        """
        if self.db_status in AssignmentState.final_unit():
            return self.db_status
        row = self.db.get_unit(self.db_id)
        assert row is not None, f"Unit {self.db_id} stopped existing in the db..."
        return row["status"]

    def set_db_status(self, status: str) -> None:
        """
        Set the status reflected in the database for this Unit
        """
        assert (
            status in AssignmentState.valid_unit()
        ), f"{status} not valid Assignment Status, not in {AssignmentState.valid_unit()}"
        self.db_status = status
        self.db.update_unit(self.db_id, status=status)

    def get_assignment(self) -> "Assignment":
        """
        Return the assignment that this Unit is part of.
        """
        if self.__assignment is None:
            from mephisto.data_model.assignment import Assignment

            self.__assignment = Assignment(self.db, self.assignment_id)
        return self.__assignment

    def get_task_run(self) -> TaskRun:
        """
        Return the task run that this assignment is part of
        """
        if self.__task_run is None:
            if self.__assignment is not None:
                self.__task_run = self.__assignment.get_task_run()
            else:
                self.__task_run = TaskRun(self.db, self.task_run_id)
        return self.__task_run

    def get_task(self) -> Task:
        """
        Return the task that this assignment is part of
        """
        if self.__task is None:
            if self.__assignment is not None:
                self.__task = self.__assignment.get_task()
            elif self.__task_run is not None:
                self.__task = self.__task_run.get_task()
            else:
                self.__task = Task(self.db, self.task_id)
        return self.__task

    def get_requester(self) -> "Requester":
        """
        Return the requester who offered this Unit
        """
        if self.__requester is None:
            if self.__assignment is not None:
                self.__requester = self.__assignment.get_requester()
            elif self.__task_run is not None:
                self.__requester = self.__task_run.get_requester()
            else:
                self.__requester = Requester(self.db, self.requester_id)
        return self.__requester

    def clear_assigned_agent(self) -> None:
        """Clear the agent that is assigned to this unit"""
        self.db.clear_unit_agent_assignment(self.db_id)
        self.agent_id = None
        self.__agent = None

    def get_assigned_agent(self) -> Optional[Agent]:
        """
        Get the agent assigned to this Unit if there is one, else return None
        """
        # In these statuses, we know the agent isn't changing anymore, and thus will
        # not need to be re-queried
        # TODO(#97) add test to ensure this behavior/assumption holds always
        if self.db_status in AssignmentState.final_unit():
            if self.agent_id is None:
                return None
            return Agent(self.db, self.agent_id)

        # Query the database to get the most up-to-date assignment, as this can
        # change after instantiation if the Unit status isn't final
        # TODO(#101) this may not be particularly efficient
        row = self.db.get_unit(self.db_id)
        assert row is not None, f"Unit {self.db_id} stopped existing in the db..."
        agent_id = row["agent_id"]
        if agent_id is not None:
            return Agent(self.db, agent_id)
        return None

    @staticmethod
    def _register_unit(
        db: "MephistoDB",
        assignment: "Assignment",
        index: int,
        pay_amount: float,
        provider_type: str,
    ) -> "Unit":
        """
        Create an entry for this unit in the database
        """
        db_id = db.new_unit(
            assignment.task_id,
            assignment.task_run_id,
            assignment.requester_id,
            assignment.db_id,
            index,
            pay_amount,
            provider_type,
            assignment.task_type,
        )
        return Unit(db, db_id)

    def get_pay_amount(self) -> float:
        """
        Return the amount that this Unit is costing against the budget,
        calculating additional fees as relevant
        """
        return self.pay_amount

    # Children classes may need to override the following

    def get_status(self) -> str:
        """
        Get the status of this unit, as determined by whether there's
        a worker working on it at the moment, and any other possible states. Should
        return one of UNIT_STATUSES

        Accurate status is crowd-provider dependent, and thus this method should be
        defined in the child class to ensure that the local record matches
        the ground truth in the provider
        """
        from mephisto.abstractions.blueprint import AgentState

        db_status = self.db_status
        computed_status = AssignmentState.LAUNCHED

        agent = self.get_assigned_agent()
        if agent is None:
            row = self.db.get_unit(self.db_id)
            computed_status = row["status"]
        else:
            agent_status = agent.get_status()
            if agent_status == AgentState.STATUS_NONE:
                computed_status = AssignmentState.LAUNCHED
            elif agent_status in [
                    AgentState.STATUS_ACCEPTED,
                    AgentState.STATUS_ONBOARDING,
                    AgentState.STATUS_PARTNER_DISCONNECT,
                    AgentState.STATUS_WAITING,
                    AgentState.STATUS_IN_TASK,
            ]:
                computed_status = AssignmentState.ASSIGNED
            elif agent_status in [AgentState.STATUS_COMPLETED]:
                computed_status = AssignmentState.COMPLETED
            elif agent_status in [AgentState.STATUS_SOFT_REJECTED]:
                computed_status = AssignmentState.SOFT_REJECTED
            elif agent_status in [AgentState.STATUS_EXPIRED]:
                computed_status = AssignmentState.EXPIRED
            elif agent_status in [
                    AgentState.STATUS_DISCONNECT,
                    AgentState.STATUS_RETURNED,
            ]:
                computed_status = AssignmentState.ASSIGNED
            elif agent_status == AgentState.STATUS_APPROVED:
                computed_status = AssignmentState.ACCEPTED
            elif agent_status == AgentState.STATUS_REJECTED:
                computed_status = AssignmentState.REJECTED

        if computed_status != db_status:
            self.set_db_status(computed_status)

        return computed_status

    # Children classes should implement the below methods

    def launch(self, task_url: str) -> None:
        """
        Make this Unit available on the crowdsourcing vendor. Depending on
        the task type, this could mean a number of different setup steps.

        Some crowd providers require setting up a configuration for the
        very first launch, and this method should call a helper to manage
        that step if necessary.
        """
        raise NotImplementedError()

    def expire(self) -> float:
        """
        Expire this unit, removing it from being workable on the vendor.
        Return the maximum time needed to wait before we know it's taken down.
        """
        raise NotImplementedError()

    def is_expired(self) -> bool:
        """Determine if this unit is expired as according to the vendor."""
        raise NotImplementedError()

    @staticmethod
    def new(db: "MephistoDB", assignment: "Assignment", index: int,
            pay_amount: float) -> "Unit":
        """
        Create a Unit for the given assignment

        Implementation should return the result of _register_unit when sure the unit
        can be successfully created to have it put into the db.
        """
        raise NotImplementedError()