mephisto.abstractions.crowd_provider
View Source
#!/usr/bin/env python3 # Copyright (c) Meta Platforms and its affiliates. # This source code is licensed under the MIT license found in the # LICENSE file in the root directory of this source tree. from abc import ABC, abstractmethod, abstractproperty from dataclasses import dataclass, field from omegaconf import MISSING, DictConfig from mephisto.abstractions.blueprint import AgentState, SharedTaskState from mephisto.data_model.unit import Unit from mephisto.data_model.requester import Requester from mephisto.data_model.worker import Worker from mephisto.data_model.agent import Agent from typing import List, Optional, Tuple, Dict, Any, ClassVar, Type, TYPE_CHECKING if TYPE_CHECKING: from mephisto.abstractions.database import MephistoDB from mephisto.data_model.task_run import TaskRun from argparse import _ArgumentGroup as ArgumentGroup @dataclass class ProviderArgs: """Base class for arguments to configure Crowd Providers""" _provider_type: str = MISSING requester_name: str = MISSING class CrowdProvider(ABC): """ Base class that defines the required functionality for the mephisto system to be able to interface with an external crowdsourcing vendor. Implementing the methods within, as well as supplying wrapped Unit, Requester, Worker, and Agent classes should ensure support for a vendor. """ PROVIDER_TYPE = "__PROVIDER_BASE_CLASS__" UnitClass: ClassVar[Type[Unit]] = Unit RequesterClass: ClassVar[Type[Requester]] = Requester WorkerClass: ClassVar[Type[Worker]] = Worker AgentClass: ClassVar[Type[Agent]] = Agent ArgsClass: ClassVar[Type[ProviderArgs]] = ProviderArgs def __init__(self, db: "MephistoDB"): """ Crowd provider classes should keep as much of their state as possible in their non-python datastore. This way the system can work even after shutdowns, and the state of the system can be managed or observed from other processes. In order to set up a datastore, init should check to see if one is already set (using get_datastore_for_provider) and use that one if available, otherwise make a new one and register it with the database. """ self.db = db if db.has_datastore_for_provider(self.PROVIDER_TYPE): self.datastore = db.get_datastore_for_provider(self.PROVIDER_TYPE) else: self.datastore_root = db.get_db_path_for_provider(self.PROVIDER_TYPE) self.datastore = self.initialize_provider_datastore(self.datastore_root) db.set_datastore_for_provider(self.PROVIDER_TYPE, self.datastore) @classmethod def is_sandbox(cls) -> bool: """Determine if the given crowd provider is a sandbox provider""" return cls.RequesterClass.is_sandbox() @classmethod def assert_task_args(cls, args: DictConfig, shared_state: "SharedTaskState"): """ Assert that the provided arguments are valid. Should fail if a task launched with these arguments would not work """ return @classmethod @abstractmethod def get_wrapper_js_path(cls): """ Return the path to the `wrap_crowd_source.js` file for this provider to be deployed to the server """ raise NotImplementedError @abstractmethod def initialize_provider_datastore(self, storage_path: str) -> Any: """ Do whatever is required to initialize this provider insofar as setting up local or external state is required to ensure that this vendor is usable. Local data storage should be put into the given root path. This method should return the local data storage component that is required to do any object initialization, as it will be available from the MephistoDB in a db.get_provider_datastore(PROVIDER_TYPE). """ raise NotImplementedError() @abstractmethod def setup_resources_for_task_run( self, task_run: "TaskRun", args: DictConfig, shared_state: "SharedTaskState", server_url: str, ) -> None: """ Setup any required resources for managing any additional resources surrounding a specific task run. """ raise NotImplementedError() @abstractmethod def cleanup_resources_from_task_run( self, task_run: "TaskRun", server_url: str ) -> None: """ Destroy any resources set up specifically for this task run """ raise NotImplementedError() def cleanup_qualification(self, qualification_name: str) -> None: """ Remove the linked qualification from the crowdprovider if it exists """ return None
View Source
class ProviderArgs: """Base class for arguments to configure Crowd Providers""" _provider_type: str = MISSING requester_name: str = MISSING
Base class for arguments to configure Crowd Providers
View Source
class CrowdProvider(ABC): """ Base class that defines the required functionality for the mephisto system to be able to interface with an external crowdsourcing vendor. Implementing the methods within, as well as supplying wrapped Unit, Requester, Worker, and Agent classes should ensure support for a vendor. """ PROVIDER_TYPE = "__PROVIDER_BASE_CLASS__" UnitClass: ClassVar[Type[Unit]] = Unit RequesterClass: ClassVar[Type[Requester]] = Requester WorkerClass: ClassVar[Type[Worker]] = Worker AgentClass: ClassVar[Type[Agent]] = Agent ArgsClass: ClassVar[Type[ProviderArgs]] = ProviderArgs def __init__(self, db: "MephistoDB"): """ Crowd provider classes should keep as much of their state as possible in their non-python datastore. This way the system can work even after shutdowns, and the state of the system can be managed or observed from other processes. In order to set up a datastore, init should check to see if one is already set (using get_datastore_for_provider) and use that one if available, otherwise make a new one and register it with the database. """ self.db = db if db.has_datastore_for_provider(self.PROVIDER_TYPE): self.datastore = db.get_datastore_for_provider(self.PROVIDER_TYPE) else: self.datastore_root = db.get_db_path_for_provider(self.PROVIDER_TYPE) self.datastore = self.initialize_provider_datastore(self.datastore_root) db.set_datastore_for_provider(self.PROVIDER_TYPE, self.datastore) @classmethod def is_sandbox(cls) -> bool: """Determine if the given crowd provider is a sandbox provider""" return cls.RequesterClass.is_sandbox() @classmethod def assert_task_args(cls, args: DictConfig, shared_state: "SharedTaskState"): """ Assert that the provided arguments are valid. Should fail if a task launched with these arguments would not work """ return @classmethod @abstractmethod def get_wrapper_js_path(cls): """ Return the path to the `wrap_crowd_source.js` file for this provider to be deployed to the server """ raise NotImplementedError @abstractmethod def initialize_provider_datastore(self, storage_path: str) -> Any: """ Do whatever is required to initialize this provider insofar as setting up local or external state is required to ensure that this vendor is usable. Local data storage should be put into the given root path. This method should return the local data storage component that is required to do any object initialization, as it will be available from the MephistoDB in a db.get_provider_datastore(PROVIDER_TYPE). """ raise NotImplementedError() @abstractmethod def setup_resources_for_task_run( self, task_run: "TaskRun", args: DictConfig, shared_state: "SharedTaskState", server_url: str, ) -> None: """ Setup any required resources for managing any additional resources surrounding a specific task run. """ raise NotImplementedError() @abstractmethod def cleanup_resources_from_task_run( self, task_run: "TaskRun", server_url: str ) -> None: """ Destroy any resources set up specifically for this task run """ raise NotImplementedError() def cleanup_qualification(self, qualification_name: str) -> None: """ Remove the linked qualification from the crowdprovider if it exists """ return None
Base class that defines the required functionality for the mephisto system to be able to interface with an external crowdsourcing vendor.
Implementing the methods within, as well as supplying wrapped Unit, Requester, Worker, and Agent classes should ensure support for a vendor.
View Source
def __init__(self, db: "MephistoDB"): """ Crowd provider classes should keep as much of their state as possible in their non-python datastore. This way the system can work even after shutdowns, and the state of the system can be managed or observed from other processes. In order to set up a datastore, init should check to see if one is already set (using get_datastore_for_provider) and use that one if available, otherwise make a new one and register it with the database. """ self.db = db if db.has_datastore_for_provider(self.PROVIDER_TYPE): self.datastore = db.get_datastore_for_provider(self.PROVIDER_TYPE) else: self.datastore_root = db.get_db_path_for_provider(self.PROVIDER_TYPE) self.datastore = self.initialize_provider_datastore(self.datastore_root) db.set_datastore_for_provider(self.PROVIDER_TYPE, self.datastore)
Crowd provider classes should keep as much of their state as possible in their non-python datastore. This way the system can work even after shutdowns, and the state of the system can be managed or observed from other processes.
In order to set up a datastore, init should check to see if one is already set (using get_datastore_for_provider) and use that one if available, otherwise make a new one and register it with the database.
View Source
@classmethod def is_sandbox(cls) -> bool: """Determine if the given crowd provider is a sandbox provider""" return cls.RequesterClass.is_sandbox()
Determine if the given crowd provider is a sandbox provider
View Source
@classmethod def assert_task_args(cls, args: DictConfig, shared_state: "SharedTaskState"): """ Assert that the provided arguments are valid. Should fail if a task launched with these arguments would not work """ return
Assert that the provided arguments are valid. Should fail if a task launched with these arguments would not work
View Source
@classmethod @abstractmethod def get_wrapper_js_path(cls): """ Return the path to the `wrap_crowd_source.js` file for this provider to be deployed to the server """ raise NotImplementedError
Return the path to the wrap_crowd_source.js
file for this
provider to be deployed to the server
View Source
@abstractmethod def initialize_provider_datastore(self, storage_path: str) -> Any: """ Do whatever is required to initialize this provider insofar as setting up local or external state is required to ensure that this vendor is usable. Local data storage should be put into the given root path. This method should return the local data storage component that is required to do any object initialization, as it will be available from the MephistoDB in a db.get_provider_datastore(PROVIDER_TYPE). """ raise NotImplementedError()
Do whatever is required to initialize this provider insofar as setting up local or external state is required to ensure that this vendor is usable.
Local data storage should be put into the given root path.
This method should return the local data storage component that is required to do any object initialization, as it will be available from the MephistoDB in a db.get_provider_datastore(PROVIDER_TYPE).
View Source
@abstractmethod def setup_resources_for_task_run( self, task_run: "TaskRun", args: DictConfig, shared_state: "SharedTaskState", server_url: str, ) -> None: """ Setup any required resources for managing any additional resources surrounding a specific task run. """ raise NotImplementedError()
Setup any required resources for managing any additional resources surrounding a specific task run.
View Source
@abstractmethod def cleanup_resources_from_task_run( self, task_run: "TaskRun", server_url: str ) -> None: """ Destroy any resources set up specifically for this task run """ raise NotImplementedError()
Destroy any resources set up specifically for this task run
View Source
def cleanup_qualification(self, qualification_name: str) -> None: """ Remove the linked qualification from the crowdprovider if it exists """ return None
Remove the linked qualification from the crowdprovider if it exists
View Source
class Unit(MephistoDataModelComponentMixin, metaclass=MephistoDBBackedABCMeta): """ 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, _used_new_call: bool = False, ): if not _used_new_call: raise AssertionError( "Direct Unit and data model access via ...Unit(db, id) is " "now deprecated in favor of calling Unit.get(db, id). " ) 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: str = row["assignment_id"] self.unit_index: int = row["unit_index"] self.pay_amount: float = row["pay_amount"] self.agent_id: Optional[str] = row["agent_id"] self.provider_type: str = row["provider_type"] self.db_status: str = row["status"] self.task_type: str = row["task_type"] self.task_id: str = row["task_id"] self.task_run_id: str = row["task_run_id"] self.sandbox: bool = row["sandbox"] self.requester_id: str = row["requester_id"] self.worker_id: str = 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, _used_new_call: bool = False, ) -> "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) -> "InitializationData": """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()}" if status == self.db_status: return logger.debug(f"Updating status for {self} to {status}") ACTIVE_UNIT_STATUSES.labels( status=self.db_status, unit_type=INDEX_TO_TYPE_MAP[self.unit_index] ).dec() ACTIVE_UNIT_STATUSES.labels( status=status, unit_type=INDEX_TO_TYPE_MAP[self.unit_index] ).inc() self.db_status = status self.db.update_unit(self.db_id, status=status) def _mark_agent_assignment(self) -> None: """Special helper to mark the transition from LAUNCHED to ASSIGNED""" assert ( self.db_status == AssignmentState.LAUNCHED ), "can only mark LAUNCHED units" ACTIVE_UNIT_STATUSES.labels( status=AssignmentState.LAUNCHED, unit_type=INDEX_TO_TYPE_MAP[self.unit_index], ).dec() ACTIVE_UNIT_STATUSES.labels( status=AssignmentState.ASSIGNED, unit_type=INDEX_TO_TYPE_MAP[self.unit_index], ).inc() 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.get(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.get(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.get(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.get(self.db, self.requester_id) return self.__requester def clear_assigned_agent(self) -> None: """Clear the agent that is assigned to this unit""" logger.debug(f"Clearing assigned agent {self.agent_id} from {self}") self.db.clear_unit_agent_assignment(self.db_id) self.get_task_run().clear_reservation(self) 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 if self.db_status in AssignmentState.final_unit(): if self.agent_id is None: return None return Agent.get(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 unit_copy = Unit.get(self.db, self.db_id) self.agent_id = unit_copy.agent_id if self.agent_id is not None: return Agent.get(self.db, self.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, sandbox=assignment.sandbox, ) unit = Unit.get(db, db_id) ACTIVE_UNIT_STATUSES.labels( status=AssignmentState.CREATED, unit_type=INDEX_TO_TYPE_MAP[index] ).inc() logger.debug(f"Registered new unit {unit} for {assignment}.") return unit 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 def __repr__(self) -> str: return f"{self.__class__.__name__}({self.db_id}, {self.db_status})" # 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 # Expiration is a terminal state, and shouldn't be changed if db_status == AssignmentState.EXPIRED: return 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, AgentState.STATUS_TIMEOUT, ]: # Still assigned, as we expect the task launcher to explicitly # update our status to expired or to remove the agent 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()
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
Inherited Members
View Source
class Requester(MephistoDataModelComponentMixin, metaclass=MephistoDBBackedABCMeta): """ High level class representing a requester on some kind of crowd provider. Sets some default initializations, but mostly should be extended by the specific requesters for crowd providers with whatever implementation details are required to get those to work. """ ArgsClass: ClassVar[Type["RequesterArgs"]] = RequesterArgs def __init__( self, db: "MephistoDB", db_id: str, row: Optional[Mapping[str, Any]] = None, _used_new_call: bool = False, ): if not _used_new_call: raise AssertionError( "Direct Requester and data model access via Requester(db, id) is " "now deprecated in favor of calling Requester.get(db, id). " ) self.db: "MephistoDB" = db if row is None: row = db.get_requester(db_id) assert row is not None, f"Given db_id {db_id} did not exist in given db" self.db_id: str = row["requester_id"] self.provider_type: str = row["provider_type"] self.requester_name: str = row["requester_name"] def __new__( cls, db: "MephistoDB", db_id: str, row: Optional[Mapping[str, Any]] = None, _used_new_call: bool = False, ) -> "Requester": """ The new method is overridden to be able to automatically generate the expected Requester class without needing to specifically find it for a given db_id. As such it is impossible to create a base Requester as you will instead be returned the correct Requester class according to the crowdprovider associated with this Requester. """ from mephisto.operations.registry import get_crowd_provider_from_type if cls == Requester: # We are trying to construct a Requester, find what type to use and # create that instead if row is None: row = db.get_requester(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"] ).RequesterClass return super().__new__(correct_class) else: # We are constructing another instance directly return super().__new__(cls) def get_task_runs(self) -> List["TaskRun"]: """ Return the list of task runs that are run by this requester """ return self.db.find_task_runs(requester_id=self.db_id) def get_total_spend(self) -> float: """ Return the total amount of funding spent by this requester across all tasks. """ task_runs = self.db.find_task_runs(requester_id=self.db_id) total_spend = 0.0 for run in task_runs: total_spend += run.get_total_spend() return total_spend @classmethod def is_sandbox(self) -> bool: """ Determine if this is a requester on a sandbox/test account """ return False def __repr__(self) -> str: return f"{self.__class__.__name__}({self.db_id})" @staticmethod def _register_requester( db: "MephistoDB", requester_id: str, provider_type: str ) -> "Requester": """ Create an entry for this requester in the database """ db_id = db.new_requester(requester_id, provider_type) requester = Requester.get(db, db_id) logger.debug(f"Registered new requester {requester}") return requester # Children classes should implement the following methods def register(self, args: Optional[DictConfig] = None) -> None: """ Register this requester with the crowd provider by providing any required credentials or such. If no args are provided, assume the registration is already made and try to assert it as such. Should throw an exception if something is wrong with provided registration arguments. """ raise NotImplementedError() def is_registered(self) -> bool: """Check to see if this requester has already been registered""" raise NotImplementedError() def get_available_budget(self) -> float: """ Return the funds that this requester profile has available for usage with its crowdsourcing vendor """ raise NotImplementedError() def to_dict(self) -> Dict[str, Any]: """ Produce a dict of this requester and important features for json serialization """ return { "requester_id": self.db_id, "provider_type": self.provider_type, "requester_name": self.requester_name, "registered": self.is_registered(), } @staticmethod def new(db: "MephistoDB", requester_name: str) -> "Requester": """ Try to create a new requester by this name, raise an exception if the name already exists. Implementation should call _register_requester(db, requester_id) when sure the requester can be successfully created to have it put into the db and return the result. """ raise NotImplementedError()
High level class representing a requester on some kind of crowd provider. Sets some default initializations, but mostly should be extended by the specific requesters for crowd providers with whatever implementation details are required to get those to work.
View Source
class Worker(MephistoDataModelComponentMixin, metaclass=MephistoDBBackedABCMeta): """ This class represents an individual - namely a person. It maintains components of ongoing identity for a user. """ def __init__( self, db: "MephistoDB", db_id: str, row: Optional[Mapping[str, Any]] = None, _used_new_call: bool = False, ): if not _used_new_call: raise AssertionError( "Direct Worker and data model access via ...Worker(db, id) is " "now deprecated in favor of calling Worker.get(db, id). " ) self.db: "MephistoDB" = db if row is None: row = db.get_worker(db_id) assert row is not None, f"Given db_id {db_id} did not exist in given db" self.db_id: str = row["worker_id"] self.provider_type = row["provider_type"] self.worker_name = row["worker_name"] # TODO(#568) Do we want any other attributes here? def __new__( cls, db: "MephistoDB", db_id: str, row: Optional[Mapping[str, Any]] = None, _used_new_call: bool = False, ) -> "Worker": """ The new method is overridden to be able to automatically generate the expected Worker class without needing to specifically find it for a given db_id. As such it is impossible to create a base Worker as you will instead be returned the correct Worker class according to the crowdprovider associated with this Worker. """ from mephisto.operations.registry import get_crowd_provider_from_type if cls == Worker: # We are trying to construct a Worker, find what type to use and # create that instead if row is None: row = db.get_worker(db_id) assert row is not None, f"Given db_id {db_id} did not exist in given db" correct_class: Type[Worker] = get_crowd_provider_from_type( row["provider_type"] ).WorkerClass return super().__new__(correct_class) else: # We are constructing another instance directly return super().__new__(cls) # TODO(#568) make getters for helpful worker statistics def get_agents(self, status: Optional[str] = None) -> List["Agent"]: """ Get the list of agents that this worker was responsible for, by the given status if needed """ assert status is None or status in AgentState.valid(), "Invalid agent status" return self.db.find_agents(worker_id=self.db_id, status=status) @staticmethod def _register_worker( db: "MephistoDB", worker_name: str, provider_type: str ) -> "Worker": """ Create an entry for this worker in the database """ db_id = db.new_worker(worker_name, provider_type) worker = Worker.get(db, db_id) logger.debug(f"Registered new worker {worker}") return worker @classmethod def new_from_provider_data( cls, db: "MephistoDB", creation_data: Dict[str, Any] ) -> "Worker": """ Given the parameters passed through wrap_crowd_source.js, construct a new worker Basic case simply takes the worker id and registers it """ return cls.new(db, creation_data["worker_name"]) def get_granted_qualification( self, qualification_name: str ) -> Optional["GrantedQualification"]: """Return the granted qualification for this worker for the given name""" found_qualifications = self.db.find_qualifications(qualification_name) if len(found_qualifications) == 0: return None qualification = found_qualifications[0] granted_qualifications = self.db.check_granted_qualifications( qualification.db_id, self.db_id ) if len(granted_qualifications) == 0: return None return granted_qualifications[0] def is_disqualified(self, qualification_name: str): """ Find out if the given worker has been disqualified by the given qualification Returns True if the qualification exists and has a falsey value Returns False if the qualification doesn't exist or has a truthy value """ qualification = self.get_granted_qualification(qualification_name) if qualification is None: return False return not qualification.value def is_qualified(self, qualification_name: str): """ Find out if the given worker has qualified by the given qualification Returns True if the qualification exists and is truthy value Returns False if the qualification doesn't exist or falsey value """ qualification = self.get_granted_qualification(qualification_name) if qualification is None: return False return bool(qualification.value) def revoke_qualification(self, qualification_name) -> bool: """ Remove this user's qualification if it exists Returns True if removal happens locally and externally, False if an exception happens with the crowd provider """ granted_qualification = self.get_granted_qualification(qualification_name) if granted_qualification is None: return False logger.debug(f"Revoking qualification {qualification_name} from worker {self}.") self.db.revoke_qualification(granted_qualification.qualification_id, self.db_id) try: self.revoke_crowd_qualification(qualification_name) return True except Exception as e: logger.exception( f"Found error while trying to revoke qualification: {repr(e)}", exc_info=True, ) return False return True def grant_qualification( self, qualification_name: str, value: int = 1, skip_crowd=False ): """ Grant a positive or negative qualification to this worker Returns True if granting happens locally and externally, False if an exception happens with the crowd provider """ found_qualifications = self.db.find_qualifications(qualification_name) if len(found_qualifications) == 0: raise Exception( f"No qualification by the name {qualification_name} found in the db" ) logger.debug( f"Granting worker {self} qualification {qualification_name}: {value}" ) qualification = found_qualifications[0] self.db.grant_qualification(qualification.db_id, self.db_id, value=value) if not skip_crowd: try: self.grant_crowd_qualification(qualification_name, value) return True except Exception as e: logger.exception( f"Found error while trying to grant qualification: {repr(e)}", exc_info=True, ) return False def __repr__(self) -> str: return f"{self.__class__.__name__}({self.db_id})" # Children classes can implement the following methods def grant_crowd_qualification( self, qualification_name: str, value: int = 1 ) -> None: """ Grant a qualification by the given name to this worker If the CrowdProvider has a notion of qualifications, they can be granted in sync with Mephisto's qualifications """ return None def revoke_crowd_qualification(self, qualification_name: str) -> None: """ Revoke the qualification by the given name from this worker If the CrowdProvider has a notion of qualifications, they can be revoked in sync with Mephisto's qualifications """ return None # Children classes should implement the following methods def bonus_worker( self, amount: float, reason: str, unit: Optional["Unit"] = None ) -> Tuple[bool, str]: """Bonus this worker for work any reason. Return success of bonus""" raise NotImplementedError() def block_worker( self, reason: str, unit: Optional["Unit"] = None, requester: Optional["Requester"] = None, ) -> Tuple[bool, str]: """Block this worker for a specified reason. Return success of block""" raise NotImplementedError() def unblock_worker(self, reason: str, requester: "Requester") -> bool: """unblock a blocked worker for the specified reason""" raise NotImplementedError() def is_blocked(self, requester: "Requester") -> bool: """Determine if a worker is blocked""" raise NotImplementedError() def is_eligible(self, task_run: "TaskRun") -> bool: """Determine if this worker is eligible for the given task run""" raise NotImplementedError() def register(self, args: Optional[Dict[str, str]] = None) -> None: """Register this worker with the crowdprovider, if necessary""" pass @staticmethod def new(db: "MephistoDB", worker_name: str) -> "Worker": """ Create a new worker attached to the given identifier, assuming it doesn't already exist in the database. Implementation should return the result of _register_worker when sure the worker can be successfully created to have it put into the db. """ raise NotImplementedError()
This class represents an individual - namely a person. It maintains components of ongoing identity for a user.
Inherited Members
- mephisto.data_model.worker.Worker
- Worker
- get_agents
- new_from_provider_data
- get_granted_qualification
- is_disqualified
- is_qualified
- revoke_qualification
- grant_qualification
- grant_crowd_qualification
- revoke_crowd_qualification
- bonus_worker
- block_worker
- unblock_worker
- is_blocked
- is_eligible
- register
- new
View Source
class Agent(MephistoDataModelComponentMixin, metaclass=MephistoDBBackedABCMeta): """ 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, _used_new_call: bool = False, ): if not _used_new_call: raise AssertionError( "Direct Agent and data model access via ...Agent(db, id) was " "now deprecated in favor of calling Agent.get(db, id). " ) 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: str = row["status"] self.worker_id: str = row["worker_id"] self.unit_id: str = row["unit_id"] self.task_type: str = row["task_type"] self.provider_type: str = row["provider_type"] self.pending_actions: "Queue[Dict[str, Any]]" = Queue() self.has_live_update = threading.Event() self.has_live_update.clear() self.assignment_id: str = row["assignment_id"] self.task_run_id: str = row["task_run_id"] self.task_id: str = row["task_id"] self.did_submit = threading.Event() self.is_shutdown = False # 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 # Related entity set by a live run self._associated_live_run: Optional["LiveTaskRun"] = None # Follow-up initialization is deferred self._state = None # type: ignore @property def state(self) -> "AgentState": if self._state is None: self._state = AgentState(self) # type: ignore return cast("AgentState", self._state) def __new__( cls, db: "MephistoDB", db_id: str, row: Optional[Mapping[str, Any]] = None, _used_new_call: bool = False, ) -> "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.operations.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 set_live_run(self, live_run: "LiveTaskRun") -> None: """Set an associated live run for this agent""" self._associated_live_run = live_run def get_live_run(self) -> "LiveTaskRun": """Return the associated live run for this agent. Throw if not set""" if self._associated_live_run is None: raise AssertionError( "Should not be getting the live run, not set for given agent" ) return self._associated_live_run 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.get(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.unit import Unit self._unit = Unit.get(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.get(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_run import TaskRun self._task_run = TaskRun.get(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.get(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 logger.debug(f"Updating {self} to {new_status}") if self.db_status in AgentState.complete(): logger.info(f"Updating {self} from final status to {new_status}") old_status = self.db_status self.db.update_agent(self.db_id, status=new_status) self.db_status = new_status if self._associated_live_run is not None: live_run = self.get_live_run() live_run.loop_wrap.execute_coro( live_run.worker_pool.push_status_update(self) ) if new_status in [ AgentState.STATUS_RETURNED, AgentState.STATUS_DISCONNECT, AgentState.STATUS_TIMEOUT, ]: # Disconnect statuses should free any pending acts self.has_live_update.set() self.did_submit.set() if old_status == AgentState.STATUS_WAITING: # Waiting agents' unit can be reassigned, as no work # has been done yet. unit = self.get_unit() logger.debug(f"Clearing {self} from {unit} for update to {new_status}") unit.clear_assigned_agent() # Metrics changes ACTIVE_AGENT_STATUSES.labels(status=old_status, agent_type="main").dec() ACTIVE_AGENT_STATUSES.labels(status=new_status, agent_type="main").inc() if ( old_status not in AgentState.complete() and new_status in AgentState.complete() ): ACTIVE_WORKERS.labels(worker_id=self.worker_id, agent_type="main").dec() @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 """ unit._mark_agent_assignment() 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.get(db, db_id) ACTIVE_AGENT_STATUSES.labels( status=AgentState.STATUS_NONE, agent_type="main" ).inc() ACTIVE_WORKERS.labels(worker_id=worker.db_id, agent_type="main").inc() logger.debug(f"Registered new agent {a} for {unit}.") 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, live_update: "Dict[str, Any]") -> None: """ Pass the observed information to the AgentState, then queue the information to be pushed to the user """ if live_update.get("update_id") is None: live_update["update_id"] = str(uuid4()) self.state.update_data(live_update) if self._associated_live_run is not None: live_run = self.get_live_run() live_run.client_io.send_live_update(self.get_agent_id(), live_update) def get_live_update( self, timeout: Optional[int] = None ) -> Optional[Dict[str, Any]]: """ 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 self.pending_actions.empty(): if timeout is None or timeout == 0: return None self.has_live_update.wait(timeout) if self.pending_actions.empty(): if self.is_shutdown: raise AgentShutdownError(self.db_id) # 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 ( not self.pending_actions.empty() ), "has_live_update released without an action!" act = self.pending_actions.get() if self.pending_actions.empty(): self.has_live_update.clear() self.state.update_data(act) return act def act(self, timeout: Optional[int] = None) -> Optional[Dict[str, Any]]: """ Request information from the Agent's frontend. If non-blocking, (timeout is None) should return None if no actions are ready to be returned. """ warn_once( "As of Mephisto 1.0 Agent.act is being deprecated in favor of Agent.get_live_update. " "This functionality will no longer work in 1.1" ) return self.get_live_update(timeout) def await_submit(self, timeout: Optional[int] = None) -> bool: """Blocking wait for this agent to submit their task""" if timeout is not None: self.did_submit.wait(timeout=timeout) return self.did_submit.is_set() def handle_submit(self, submit_data: Dict[str, Any]) -> None: """Handle final submission for an onboarding agent, with the given data""" self.did_submit.set() self.state.update_submit(submit_data) 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_live_update.set() if self._associated_live_run is not None: live_run = self.get_live_run() live_run.loop_wrap.execute_coro( live_run.worker_pool.push_status_update(self) ) self.db_status = row["status"] return self.db_status def shutdown(self) -> None: """ Force the given agent to end any polling threads and throw an AgentShutdownError from any acts called on it, ensuring tasks using this agent can be cleaned up. """ logger.debug(f"{self} is shutting down") self.has_live_update.set() self.is_shutdown = True def __repr__(self) -> str: return f"{self.__class__.__name__}({self.db_id}, {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()
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.
Inherited Members
- mephisto.data_model.agent.Agent
- Agent
- state
- set_live_run
- get_live_run
- get_agent_id
- get_worker
- get_unit
- get_assignment
- get_task_run
- get_task
- get_data_dir
- update_status
- new_from_provider_data
- observe
- get_live_update
- act
- await_submit
- handle_submit
- get_status
- shutdown
- approve_work
- soft_reject_work
- reject_work
- mark_done
- new
View Source
class ProviderArgs: """Base class for arguments to configure Crowd Providers""" _provider_type: str = MISSING requester_name: str = MISSING
Base class for arguments to configure Crowd Providers