This documentation is for an older version (1.4.7) of Dagster. You can view the version of this page from our latest release below.
from abc import ABC, abstractmethod
from typing import (
AbstractSet,
Any,
Dict,
Iterator,
List,
Mapping,
Optional,
Sequence,
Set,
cast,
)
from typing_extensions import TypeAlias
import dagster._check as check
from dagster._annotations import deprecated, experimental, public
from dagster._core.definitions.assets import AssetsDefinition
from dagster._core.definitions.data_version import (
DataProvenance,
DataVersion,
extract_data_provenance_from_entry,
)
from dagster._core.definitions.dependency import Node, NodeHandle
from dagster._core.definitions.events import (
AssetKey,
AssetMaterialization,
AssetObservation,
ExpectationResult,
UserEvent,
)
from dagster._core.definitions.job_definition import JobDefinition
from dagster._core.definitions.op_definition import OpDefinition
from dagster._core.definitions.partition import PartitionsDefinition
from dagster._core.definitions.partition_key_range import PartitionKeyRange
from dagster._core.definitions.step_launcher import StepLauncher
from dagster._core.definitions.time_window_partitions import TimeWindow
from dagster._core.errors import (
DagsterInvalidPropertyError,
DagsterInvariantViolationError,
)
from dagster._core.events import DagsterEvent
from dagster._core.instance import DagsterInstance
from dagster._core.log_manager import DagsterLogManager
from dagster._core.storage.dagster_run import DagsterRun
from dagster._utils.forked_pdb import ForkedPdb
from .system import StepExecutionContext
class AbstractComputeExecutionContext(ABC):
"""Base class for op context implemented by OpExecutionContext and DagstermillExecutionContext."""
@abstractmethod
def has_tag(self, key: str) -> bool:
"""Implement this method to check if a logging tag is set."""
@abstractmethod
def get_tag(self, key: str) -> Optional[str]:
"""Implement this method to get a logging tag."""
@property
@abstractmethod
def run_id(self) -> str:
"""The run id for the context."""
@property
@abstractmethod
def op_def(self) -> OpDefinition:
"""The op definition corresponding to the execution step being executed."""
@property
@abstractmethod
def job_def(self) -> JobDefinition:
"""The job being executed."""
@property
@abstractmethod
def run(self) -> DagsterRun:
"""The DagsterRun object corresponding to the execution."""
@property
@abstractmethod
def resources(self) -> Any:
"""Resources available in the execution context."""
@property
@abstractmethod
def log(self) -> DagsterLogManager:
"""The log manager available in the execution context."""
@property
@abstractmethod
def op_config(self) -> Any:
"""The parsed config specific to this op."""
[docs]class OpExecutionContext(AbstractComputeExecutionContext):
"""The ``context`` object that can be made available as the first argument to the function
used for computing an op or asset.
This context object provides system information such as resources, config, and logging.
To construct an execution context for testing purposes, use :py:func:`dagster.build_op_context`.
Example:
.. code-block:: python
from dagster import op, OpExecutionContext
@op
def hello_world(context: OpExecutionContext):
context.log.info("Hello, world!")
"""
__slots__ = ["_step_execution_context"]
def __init__(self, step_execution_context: StepExecutionContext):
self._step_execution_context = check.inst_param(
step_execution_context,
"step_execution_context",
StepExecutionContext,
)
self._pdb: Optional[ForkedPdb] = None
self._events: List[DagsterEvent] = []
self._output_metadata: Dict[str, Any] = {}
@public
@property
def op_config(self) -> Any:
"""Any: The parsed config specific to this op."""
return self._step_execution_context.op_config
@property
def dagster_run(self) -> DagsterRun:
"""PipelineRun: The current pipeline run."""
return self._step_execution_context.dagster_run
@property
def run(self) -> DagsterRun:
"""DagsterRun: The current run."""
return self.dagster_run
@public
@property
def instance(self) -> DagsterInstance:
"""DagsterInstance: The current Dagster instance."""
return self._step_execution_context.instance
@public
@property
def pdb(self) -> ForkedPdb:
"""dagster.utils.forked_pdb.ForkedPdb: Gives access to pdb debugging from within the op.
Example:
.. code-block:: python
@op
def debug(context):
context.pdb.set_trace()
"""
if self._pdb is None:
self._pdb = ForkedPdb()
return self._pdb
@property
def file_manager(self):
"""Deprecated access to the file manager.
:meta private:
"""
raise DagsterInvalidPropertyError(
"You have attempted to access the file manager which has been moved to resources in"
" 0.10.0. Please access it via `context.resources.file_manager` instead."
)
@public
@property
def resources(self) -> Any:
"""Resources: The currently available resources."""
return self._step_execution_context.resources
@property
def step_launcher(self) -> Optional[StepLauncher]:
"""Optional[StepLauncher]: The current step launcher, if any."""
return self._step_execution_context.step_launcher
@public
@property
def run_id(self) -> str:
"""str: The id of the current execution's run."""
return self._step_execution_context.run_id
@public
@property
def run_config(self) -> Mapping[str, object]:
"""dict: The run config for the current execution."""
return self._step_execution_context.run_config
@public
@property
def job_def(self) -> JobDefinition:
"""JobDefinition: The currently executing pipeline."""
return self._step_execution_context.job_def
@public
@property
def job_name(self) -> str:
"""str: The name of the currently executing pipeline."""
return self._step_execution_context.job_name
@public
@property
def log(self) -> DagsterLogManager:
"""DagsterLogManager: The log manager available in the execution context."""
return self._step_execution_context.log
@property
def node_handle(self) -> NodeHandle:
"""NodeHandle: The current op's handle.
:meta private:
"""
return self._step_execution_context.node_handle
@property
def op_handle(self) -> NodeHandle:
"""NodeHandle: The current op's handle.
:meta private:
"""
return self.node_handle
@property
def op(self) -> Node:
"""Node: The object representing the invoked op within the graph.
:meta private:
"""
return self._step_execution_context.job_def.get_node(self.node_handle)
@public
@property
def op_def(self) -> OpDefinition:
"""OpDefinition: The current op definition."""
return cast(OpDefinition, self.op.definition)
@public
@property
def has_partition_key(self) -> bool:
"""Whether the current run is a partitioned run."""
return self._step_execution_context.has_partition_key
@public
@property
def partition_key(self) -> str:
"""The partition key for the current run.
Raises an error if the current run is not a partitioned run.
"""
return self._step_execution_context.partition_key
@deprecated(breaking_version="2.0", additional_warn_text="Use `partition_key_range` instead.")
@public
@property
def asset_partition_key_range(self) -> PartitionKeyRange:
"""The range of partition keys for the current run.
If run is for a single partition key, return a `PartitionKeyRange` with the same start and
end. Raises an error if the current run is not a partitioned run.
"""
return self.partition_key_range
@public
@property
def partition_key_range(self) -> PartitionKeyRange:
"""The range of partition keys for the current run.
If run is for a single partition key, return a `PartitionKeyRange` with the same start and
end. Raises an error if the current run is not a partitioned run.
"""
return self._step_execution_context.asset_partition_key_range
@public
@property
def partition_time_window(self) -> TimeWindow:
"""The partition time window for the current run.
Raises an error if the current run is not a partitioned run, or if the job's partition
definition is not a TimeWindowPartitionsDefinition.
"""
return self._step_execution_context.partition_time_window
[docs] @public
def has_tag(self, key: str) -> bool:
"""Check if a logging tag is set.
Args:
key (str): The tag to check.
Returns:
bool: Whether the tag is set.
"""
return self._step_execution_context.has_tag(key)
[docs] @public
def get_tag(self, key: str) -> Optional[str]:
"""Get a logging tag.
Args:
key (tag): The tag to get.
Returns:
Optional[str]: The value of the tag, if present.
"""
return self._step_execution_context.get_tag(key)
@property
def run_tags(self) -> Mapping[str, str]:
"""Mapping[str, str]: The tags for the current run."""
return self._step_execution_context.run_tags
def has_events(self) -> bool:
return bool(self._events)
def consume_events(self) -> Iterator[DagsterEvent]:
"""Pops and yields all user-generated events that have been recorded from this context.
If consume_events has not yet been called, this will yield all logged events since the beginning of the op's computation. If consume_events has been called, it will yield all events since the last time consume_events was called. Designed for internal use. Users should never need to invoke this method.
"""
events = self._events
self._events = []
yield from events
[docs] @public
def log_event(self, event: UserEvent) -> None:
"""Log an AssetMaterialization, AssetObservation, or ExpectationResult from within the body of an op.
Events logged with this method will appear in the list of DagsterEvents, as well as the event log.
Args:
event (Union[AssetMaterialization, AssetObservation, ExpectationResult]): The event to log.
**Examples:**
.. code-block:: python
from dagster import op, AssetMaterialization
@op
def log_materialization(context):
context.log_event(AssetMaterialization("foo"))
"""
if isinstance(event, AssetMaterialization):
self._events.append(
DagsterEvent.asset_materialization(self._step_execution_context, event)
)
elif isinstance(event, AssetObservation):
self._events.append(DagsterEvent.asset_observation(self._step_execution_context, event))
elif isinstance(event, ExpectationResult):
self._events.append(
DagsterEvent.step_expectation_result(self._step_execution_context, event)
)
else:
check.failed(f"Unexpected event {event}")
def add_output_metadata(
self,
metadata: Mapping[str, Any],
output_name: Optional[str] = None,
mapping_key: Optional[str] = None,
) -> None:
"""Add metadata to one of the outputs of an op.
This can only be used once per output in the body of an op. Using this method with the same output_name more than once within an op will result in an error.
Args:
metadata (Mapping[str, Any]): The metadata to attach to the output
output_name (Optional[str]): The name of the output to attach metadata to. If there is only one output on the op, then this argument does not need to be provided. The metadata will automatically be attached to the only output.
**Examples:**
.. code-block:: python
from dagster import Out, op
from typing import Tuple
@op
def add_metadata(context):
context.add_output_metadata({"foo", "bar"})
return 5 # Since the default output is called "result", metadata will be attached to the output "result".
@op(out={"a": Out(), "b": Out()})
def add_metadata_two_outputs(context) -> Tuple[str, int]:
context.add_output_metadata({"foo": "bar"}, output_name="b")
context.add_output_metadata({"baz": "bat"}, output_name="a")
return ("dog", 5)
"""
metadata = check.mapping_param(metadata, "metadata", key_type=str)
output_name = check.opt_str_param(output_name, "output_name")
mapping_key = check.opt_str_param(mapping_key, "mapping_key")
self._step_execution_context.add_output_metadata(
metadata=metadata, output_name=output_name, mapping_key=mapping_key
)
def get_output_metadata(
self, output_name: str, mapping_key: Optional[str] = None
) -> Optional[Mapping[str, Any]]:
return self._step_execution_context.get_output_metadata(
output_name=output_name, mapping_key=mapping_key
)
def get_step_execution_context(self) -> StepExecutionContext:
"""Allows advanced users (e.g. framework authors) to punch through to the underlying
step execution context.
:meta private:
Returns:
StepExecutionContext: The underlying system context.
"""
return self._step_execution_context
@public
@property
def retry_number(self) -> int:
"""Which retry attempt is currently executing i.e. 0 for initial attempt, 1 for first retry, etc."""
return self._step_execution_context.previous_attempt_count
def describe_op(self):
return self._step_execution_context.describe_op()
[docs] @public
def get_mapping_key(self) -> Optional[str]:
"""Which mapping_key this execution is for if downstream of a DynamicOutput, otherwise None."""
return self._step_execution_context.step.get_mapping_key()
#############################################################################################
# asset related methods
#############################################################################################
@public
@property
def has_assets_def(self) -> bool:
"""If there is a backing AssetsDefinition for what is currently executing."""
assets_def = self.job_def.asset_layer.assets_def_for_node(self.node_handle)
return assets_def is not None
@public
@property
def assets_def(self) -> AssetsDefinition:
"""The backing AssetsDefinition for what is currently executing, errors if not available."""
assets_def = self.job_def.asset_layer.assets_def_for_node(self.node_handle)
if assets_def is None:
raise DagsterInvalidPropertyError(
f"Op '{self.op.name}' does not have an assets definition."
)
return assets_def
@public
@property
def selected_asset_keys(self) -> AbstractSet[AssetKey]:
"""Get the set of AssetKeys this execution is expected to materialize."""
assets_def = self.job_def.asset_layer.assets_def_for_node(self.node_handle)
if assets_def is None:
return set()
return assets_def.keys
@public
@property
def selected_output_names(self) -> AbstractSet[str]:
"""Get the output names that correspond to the current selection of assets this execution is expected to materialize."""
# map selected asset keys to the output names they correspond to
selected_asset_keys = self.selected_asset_keys
selected_outputs: Set[str] = set()
for output_name in self.op.output_dict.keys():
asset_info = self.job_def.asset_layer.asset_info_for_output(
self.node_handle, output_name
)
if any( # For graph-backed assets, check if a downstream asset is selected
[
asset_key in selected_asset_keys
for asset_key in self.job_def.asset_layer.downstream_dep_assets(
self.node_handle, output_name
)
]
) or (asset_info and asset_info.key in selected_asset_keys):
selected_outputs.add(output_name)
return selected_outputs
[docs] @public
def asset_key_for_output(self, output_name: str = "result") -> AssetKey:
"""Return the AssetKey for the corresponding output."""
asset_output_info = self.job_def.asset_layer.asset_info_for_output(
node_handle=self.op_handle, output_name=output_name
)
if asset_output_info is None:
check.failed(f"Output '{output_name}' has no asset")
else:
return asset_output_info.key
[docs] @public
def output_for_asset_key(self, asset_key: AssetKey) -> str:
"""Return the output name for the corresponding asset key."""
node_output_handle = self.job_def.asset_layer.node_output_handle_for_asset(asset_key)
if node_output_handle is None:
check.failed(f"Asset key '{asset_key}' has no output")
else:
return node_output_handle.output_name
[docs] @public
def asset_key_for_input(self, input_name: str) -> AssetKey:
"""Return the AssetKey for the corresponding input."""
key = self.job_def.asset_layer.asset_key_for_input(
node_handle=self.op_handle, input_name=input_name
)
if key is None:
check.failed(f"Input '{input_name}' has no asset")
else:
return key
[docs] @public
def asset_partition_key_for_output(self, output_name: str = "result") -> str:
"""Returns the asset partition key for the given output. Defaults to "result", which is the
name of the default output.
"""
return self._step_execution_context.asset_partition_key_for_output(output_name)
[docs] @public
def asset_partitions_time_window_for_output(self, output_name: str = "result") -> TimeWindow:
"""The time window for the partitions of the output asset.
Raises an error if either of the following are true:
- The output asset has no partitioning.
- The output asset is not partitioned with a TimeWindowPartitionsDefinition or a
MultiPartitionsDefinition with one time-partitioned dimension.
"""
return self._step_execution_context.asset_partitions_time_window_for_output(output_name)
[docs] @public
def asset_partition_key_range_for_output(
self, output_name: str = "result"
) -> PartitionKeyRange:
"""Return the PartitionKeyRange for the corresponding output. Errors if not present."""
return self._step_execution_context.asset_partition_key_range_for_output(output_name)
[docs] @public
def asset_partition_key_range_for_input(self, input_name: str) -> PartitionKeyRange:
"""Return the PartitionKeyRange for the corresponding input. Errors if there is more or less than one."""
return self._step_execution_context.asset_partition_key_range_for_input(input_name)
[docs] @public
def asset_partition_key_for_input(self, input_name: str) -> str:
"""Returns the partition key of the upstream asset corresponding to the given input."""
return self._step_execution_context.asset_partition_key_for_input(input_name)
[docs] @public
def asset_partitions_def_for_output(self, output_name: str = "result") -> PartitionsDefinition:
"""The PartitionsDefinition on the upstream asset corresponding to this input."""
asset_key = self.asset_key_for_output(output_name)
result = self._step_execution_context.job_def.asset_layer.partitions_def_for_asset(
asset_key
)
if result is None:
raise DagsterInvariantViolationError(
f"Attempting to access partitions def for asset {asset_key}, but it is not"
" partitioned"
)
return result
[docs] @public
def asset_partitions_def_for_input(self, input_name: str) -> PartitionsDefinition:
"""The PartitionsDefinition on the upstream asset corresponding to this input."""
asset_key = self.asset_key_for_input(input_name)
result = self._step_execution_context.job_def.asset_layer.partitions_def_for_asset(
asset_key
)
if result is None:
raise DagsterInvariantViolationError(
f"Attempting to access partitions def for asset {asset_key}, but it is not"
" partitioned"
)
return result
[docs] @public
def asset_partition_keys_for_output(self, output_name: str = "result") -> Sequence[str]:
"""Returns a list of the partition keys for the given output."""
return self.asset_partitions_def_for_output(output_name).get_partition_keys_in_range(
self._step_execution_context.asset_partition_key_range_for_output(output_name),
dynamic_partitions_store=self.instance,
)
[docs] @public
def asset_partition_keys_for_input(self, input_name: str) -> Sequence[str]:
"""Returns a list of the partition keys of the upstream asset corresponding to the
given input.
"""
return list(
self._step_execution_context.asset_partitions_subset_for_input(
input_name
).get_partition_keys()
)
[docs] @public
def asset_partitions_time_window_for_input(self, input_name: str = "result") -> TimeWindow:
"""The time window for the partitions of the input asset.
Raises an error if either of the following are true:
- The input asset has no partitioning.
- The input asset is not partitioned with a TimeWindowPartitionsDefinition or a
MultiPartitionsDefinition with one time-partitioned dimension.
"""
return self._step_execution_context.asset_partitions_time_window_for_input(input_name)
[docs] @public
@experimental
def get_asset_provenance(self, asset_key: AssetKey) -> Optional[DataProvenance]:
"""Return the provenance information for the most recent materialization of an asset.
Args:
asset_key (AssetKey): Key of the asset for which to retrieve provenance.
Returns:
Optional[DataProvenance]: Provenance information for the most recent
materialization of the asset. Returns `None` if the asset was never materialized or
the materialization record is too old to contain provenance information.
"""
record = self.instance.get_latest_data_version_record(asset_key)
return (
None if record is None else extract_data_provenance_from_entry(record.event_log_entry)
)
def set_data_version(self, asset_key: AssetKey, data_version: DataVersion) -> None:
"""Set the data version for an asset being materialized by the currently executing step.
This is useful for external execution situations where it is not possible to return
an `Output`.
Args:
asset_key (AssetKey): Key of the asset for which to set the data version.
data_version (DataVersion): The data version to set.
"""
self._step_execution_context.set_data_version(asset_key, data_version)
# actually forking the object type for assets is tricky for users in the cases of:
# * manually constructing ops to make AssetsDefinitions
# * having ops in a graph that form a graph backed asset
# so we have a single type that users can call by their preferred name where appropriate
AssetExecutionContext: TypeAlias = OpExecutionContext