from enum import Enum
from typing import (
TYPE_CHECKING,
AbstractSet,
Any,
Callable,
Generic,
List,
Mapping,
NamedTuple,
Optional,
Sequence,
TypeVar,
Union,
cast,
)
import dagster._check as check
from dagster._annotations import PublicAttr, deprecated, experimental_param, public
from dagster._core.definitions.asset_key import (
AssetKey as AssetKey,
CoercibleToAssetKey as CoercibleToAssetKey,
CoercibleToAssetKeyPrefix as CoercibleToAssetKeyPrefix,
parse_asset_key_string,
)
from dagster._core.definitions.data_version import DATA_VERSION_TAG, DataVersion
from dagster._core.storage.tags import MULTIDIMENSIONAL_PARTITION_PREFIX, SYSTEM_TAG_PREFIX
from dagster._serdes import whitelist_for_serdes
from dagster._serdes.serdes import NamedTupleSerializer
from .metadata import (
MetadataFieldSerializer,
MetadataMapping,
MetadataValue,
RawMetadataValue,
normalize_metadata,
)
from .utils import DEFAULT_OUTPUT, check_valid_name
if TYPE_CHECKING:
from dagster._core.execution.context.output import OutputContext
class AssetKeyPartitionKey(NamedTuple):
"""An AssetKey with an (optional) partition key. Refers either to a non-partitioned asset or a
partition of a partitioned asset.
"""
asset_key: AssetKey
partition_key: Optional[str] = None
DynamicAssetKey = Callable[["OutputContext"], Optional[AssetKey]]
@whitelist_for_serdes
class AssetLineageInfo(
NamedTuple("_AssetLineageInfo", [("asset_key", AssetKey), ("partitions", AbstractSet[str])])
):
def __new__(cls, asset_key: AssetKey, partitions: Optional[AbstractSet[str]] = None):
asset_key = check.inst_param(asset_key, "asset_key", AssetKey)
partitions = check.opt_set_param(partitions, "partitions", str)
return super(AssetLineageInfo, cls).__new__(cls, asset_key=asset_key, partitions=partitions)
T = TypeVar("T")
[docs]@experimental_param(param="data_version")
class Output(Generic[T]):
"""Event corresponding to one of an op's outputs.
Op compute functions must explicitly yield events of this type when they have more than
one output, or when they also yield events of other types, or when defining a op using the
:py:class:`OpDefinition` API directly.
Outputs are values produced by ops that will be consumed by downstream ops in a job.
They are type-checked at op boundaries when their corresponding :py:class:`Out`
or the downstream :py:class:`In` is typed.
Args:
value (Any): The value returned by the compute function.
output_name (Optional[str]): Name of the corresponding out. (default:
"result")
metadata (Optional[Dict[str, Union[str, float, int, MetadataValue]]]):
Arbitrary metadata about the output. Keys are displayed string labels, and values are
one of the following: string, float, int, JSON-serializable dict, JSON-serializable
list, and one of the data classes returned by a MetadataValue static method.
data_version (Optional[DataVersion]): (Experimental) A data version to manually set
for the asset.
"""
def __init__(
self,
value: T,
output_name: Optional[str] = DEFAULT_OUTPUT,
metadata: Optional[Mapping[str, RawMetadataValue]] = None,
data_version: Optional[DataVersion] = None,
):
self._value = value
self._output_name = check.str_param(output_name, "output_name")
self._data_version = check.opt_inst_param(data_version, "data_version", DataVersion)
self._metadata = normalize_metadata(
check.opt_mapping_param(metadata, "metadata", key_type=str),
)
@property
def metadata(self) -> MetadataMapping:
return self._metadata
@public
@property
def value(self) -> Any:
"""Any: The value returned by the compute function."""
return self._value
@public
@property
def output_name(self) -> str:
"""str: Name of the corresponding :py:class:`Out`."""
return self._output_name
@public
@property
def data_version(self) -> Optional[DataVersion]:
"""Optional[DataVersion]: A data version that was manually set on the `Output`."""
return self._data_version
def __eq__(self, other: object) -> bool:
return (
isinstance(other, Output)
and self.value == other.value
and self.output_name == other.output_name
and self.metadata == other.metadata
)
[docs]class DynamicOutput(Generic[T]):
"""Variant of :py:class:`Output <dagster.Output>` used to support
dynamic mapping & collect. Each ``DynamicOutput`` produced by an op represents
one item in a set that can be processed individually with ``map`` or gathered
with ``collect``.
Each ``DynamicOutput`` must have a unique ``mapping_key`` to distinguish it with it's set.
Args:
value (Any):
The value returned by the compute function.
mapping_key (str):
The key that uniquely identifies this dynamic value relative to its peers.
This key will be used to identify the downstream ops when mapped, ie
``mapped_op[example_mapping_key]``
output_name (Optional[str]):
Name of the corresponding :py:class:`DynamicOut` defined on the op.
(default: "result")
metadata (Optional[Dict[str, Union[str, float, int, MetadataValue]]]):
Arbitrary metadata about the failure. Keys are displayed string labels, and values are
one of the following: string, float, int, JSON-serializable dict, JSON-serializable
list, and one of the data classes returned by a MetadataValue static method.
"""
def __init__(
self,
value: T,
mapping_key: str,
output_name: Optional[str] = DEFAULT_OUTPUT,
metadata: Optional[Mapping[str, RawMetadataValue]] = None,
):
self._mapping_key = check_valid_name(check.str_param(mapping_key, "mapping_key"))
self._output_name = check.str_param(output_name, "output_name")
self._value = value
self._metadata = normalize_metadata(
check.opt_mapping_param(metadata, "metadata", key_type=str),
)
@property
def metadata(self) -> Mapping[str, MetadataValue]:
return self._metadata
@public
@property
def mapping_key(self) -> str:
"""The mapping_key that was set for this DynamicOutput at instantiation."""
return self._mapping_key
@public
@property
def value(self) -> T:
"""The value that is returned by the compute function for this DynamicOut."""
return self._value
@public
@property
def output_name(self) -> str:
"""Name of the :py:class:`DynamicOut` defined on the op that this DynamicOut is associated with."""
return self._output_name
def __eq__(self, other: object) -> bool:
return (
isinstance(other, DynamicOutput)
and self.value == other.value
and self.output_name == other.output_name
and self.mapping_key == other.mapping_key
and self.metadata == other.metadata
)
[docs]@whitelist_for_serdes(
storage_field_names={"metadata": "metadata_entries"},
field_serializers={"metadata": MetadataFieldSerializer},
)
class AssetObservation(
NamedTuple(
"_AssetObservation",
[
("asset_key", PublicAttr[AssetKey]),
("description", PublicAttr[Optional[str]]),
("metadata", PublicAttr[Mapping[str, MetadataValue]]),
("partition", PublicAttr[Optional[str]]),
("tags", PublicAttr[Mapping[str, str]]),
],
)
):
"""Event that captures metadata about an asset at a point in time.
Args:
asset_key (Union[str, List[str], AssetKey]): A key to identify the asset.
partition (Optional[str]): The name of a partition of the asset that the metadata
corresponds to.
tags (Optional[Mapping[str, str]]): A mapping containing system-populated tags for the
observation. Users should not pass values into this argument.
metadata (Optional[Dict[str, Union[str, float, int, MetadataValue]]]):
Arbitrary metadata about the asset. Keys are displayed string labels, and values are
one of the following: string, float, int, JSON-serializable dict, JSON-serializable
list, and one of the data classes returned by a MetadataValue static method.
"""
def __new__(
cls,
asset_key: CoercibleToAssetKey,
description: Optional[str] = None,
metadata: Optional[Mapping[str, RawMetadataValue]] = None,
partition: Optional[str] = None,
tags: Optional[Mapping[str, str]] = None,
):
if isinstance(asset_key, AssetKey):
check.inst_param(asset_key, "asset_key", AssetKey)
elif isinstance(asset_key, str):
asset_key = AssetKey(parse_asset_key_string(asset_key))
else:
check.sequence_param(asset_key, "asset_key", of_type=str)
asset_key = AssetKey(asset_key)
tags = check.opt_mapping_param(tags, "tags", key_type=str, value_type=str)
if any([not tag.startswith(SYSTEM_TAG_PREFIX) for tag in tags or {}]):
check.failed(
"Users should not pass values into the tags argument for AssetMaterializations. "
"The tags argument is reserved for system-populated tags."
)
normed_metadata = normalize_metadata(
check.opt_mapping_param(metadata, "metadata", key_type=str),
)
return super(AssetObservation, cls).__new__(
cls,
asset_key=asset_key,
description=check.opt_str_param(description, "description"),
metadata=normed_metadata,
tags=tags,
partition=check.opt_str_param(partition, "partition"),
)
@property
def label(self) -> str:
return " ".join(self.asset_key.path)
@property
def data_version(self) -> Optional[str]:
return self.tags.get(DATA_VERSION_TAG)
UNDEFINED_ASSET_KEY_PATH = ["__undefined__"]
class AssetMaterializationSerializer(NamedTupleSerializer):
# There are old `Materialization` objects in storage. We set the default value for asset key to
# be `AssetKey(["__undefined__"])` to ensure that we can load these objects, without needing to
# allow for the construction of new `AssetMaterialization` objects with no defined AssetKey.
def before_unpack(self, context, unpacked_dict: Any) -> Any:
# cover both the case where "asset_key" is not present at all and where it is None
if unpacked_dict.get("asset_key") is None:
unpacked_dict["asset_key"] = AssetKey(UNDEFINED_ASSET_KEY_PATH)
return unpacked_dict
[docs]@whitelist_for_serdes(
old_storage_names={"Materialization"},
serializer=AssetMaterializationSerializer,
storage_field_names={"metadata": "metadata_entries"},
field_serializers={"metadata": MetadataFieldSerializer},
)
class AssetMaterialization(
NamedTuple(
"_AssetMaterialization",
[
("asset_key", PublicAttr[AssetKey]),
("description", PublicAttr[Optional[str]]),
("metadata", PublicAttr[Mapping[str, MetadataValue]]),
("partition", PublicAttr[Optional[str]]),
("tags", Optional[Mapping[str, str]]),
],
)
):
"""Event indicating that an op has materialized an asset.
Op compute functions may yield events of this type whenever they wish to indicate to the
Dagster framework (and the end user) that they have produced a materialized value as a
side effect of computation. Unlike outputs, asset materializations can not be passed to other
ops, and their persistence is controlled by op logic, rather than by the Dagster
framework.
Op authors should use these events to organize metadata about the side effects of their
computations, enabling tooling like the Assets dashboard in the Dagster UI.
Args:
asset_key (Union[str, List[str], AssetKey]): A key to identify the materialized asset across
job runs
description (Optional[str]): A longer human-readable description of the materialized value.
partition (Optional[str]): The name of the partition
that was materialized.
tags (Optional[Mapping[str, str]]): A mapping containing system-populated tags for the
materialization. Users should not pass values into this argument.
metadata (Optional[Dict[str, RawMetadataValue]]):
Arbitrary metadata about the asset. Keys are displayed string labels, and values are
one of the following: string, float, int, JSON-serializable dict, JSON-serializable
list, and one of the data classes returned by a MetadataValue static method.
"""
def __new__(
cls,
asset_key: CoercibleToAssetKey,
description: Optional[str] = None,
metadata: Optional[Mapping[str, RawMetadataValue]] = None,
partition: Optional[str] = None,
tags: Optional[Mapping[str, str]] = None,
):
from dagster._core.definitions.multi_dimensional_partitions import MultiPartitionKey
if isinstance(asset_key, AssetKey):
check.inst_param(asset_key, "asset_key", AssetKey)
elif isinstance(asset_key, str):
asset_key = AssetKey(parse_asset_key_string(asset_key))
else:
check.sequence_param(asset_key, "asset_key", of_type=str)
asset_key = AssetKey(asset_key)
check.opt_mapping_param(tags, "tags", key_type=str, value_type=str)
invalid_tags = [tag for tag in tags or {} if not tag.startswith(SYSTEM_TAG_PREFIX)]
if len(invalid_tags) > 0:
check.failed(
f"Invalid tags: {tags} Users should not pass values into the tags argument for"
" AssetMaterializations. The tags argument is reserved for system-populated tags."
)
normed_metadata = normalize_metadata(
check.opt_mapping_param(metadata, "metadata", key_type=str),
)
partition = check.opt_str_param(partition, "partition")
if not isinstance(partition, MultiPartitionKey):
# When event log records are unpacked from storage, cast the partition key as a
# MultiPartitionKey if multi-dimensional partition tags exist
multi_dimensional_partitions = {
dimension[len(MULTIDIMENSIONAL_PARTITION_PREFIX) :]: partition_key
for dimension, partition_key in (tags or {}).items()
if dimension.startswith(MULTIDIMENSIONAL_PARTITION_PREFIX)
}
if multi_dimensional_partitions:
partition = MultiPartitionKey(multi_dimensional_partitions)
return super(AssetMaterialization, cls).__new__(
cls,
asset_key=asset_key,
description=check.opt_str_param(description, "description"),
metadata=normed_metadata,
tags=tags,
partition=partition,
)
@property
def label(self) -> str:
return " ".join(self.asset_key.path)
[docs] @public
@staticmethod
def file(
path: str,
description: Optional[str] = None,
asset_key: Optional[Union[str, Sequence[str], AssetKey]] = None,
) -> "AssetMaterialization":
"""Static constructor for standard materializations corresponding to files on disk.
Args:
path (str): The path to the file.
description (Optional[str]): A human-readable description of the materialization.
"""
if not asset_key:
asset_key = path
return AssetMaterialization(
asset_key=cast(Union[str, AssetKey, List[str]], asset_key),
description=description,
metadata={"path": MetadataValue.path(path)},
)
[docs]@deprecated(
breaking_version="2.0",
additional_warn_text="If using assets, use AssetCheckResult and @asset_check instead.",
)
@whitelist_for_serdes(
storage_field_names={"metadata": "metadata_entries"},
field_serializers={"metadata": MetadataFieldSerializer},
)
class ExpectationResult(
NamedTuple(
"_ExpectationResult",
[
("success", PublicAttr[bool]),
("label", PublicAttr[Optional[str]]),
("description", PublicAttr[Optional[str]]),
("metadata", PublicAttr[Mapping[str, MetadataValue]]),
],
)
):
"""Event corresponding to a data quality test.
Op compute functions may yield events of this type whenever they wish to indicate to the
Dagster framework (and the end user) that a data quality test has produced a (positive or
negative) result.
Args:
success (bool): Whether the expectation passed or not.
label (Optional[str]): Short display name for expectation. Defaults to "result".
description (Optional[str]): A longer human-readable description of the expectation.
metadata (Optional[Dict[str, RawMetadataValue]]):
Arbitrary metadata about the failure. Keys are displayed string labels, and values are
one of the following: string, float, int, JSON-serializable dict, JSON-serializable
list, and one of the data classes returned by a MetadataValue static method.
"""
def __new__(
cls,
success: bool,
label: Optional[str] = None,
description: Optional[str] = None,
metadata: Optional[Mapping[str, RawMetadataValue]] = None,
):
normed_metadata = normalize_metadata(
check.opt_mapping_param(metadata, "metadata", key_type=str),
)
return super(ExpectationResult, cls).__new__(
cls,
success=check.bool_param(success, "success"),
label=check.opt_str_param(label, "label", "result"),
description=check.opt_str_param(description, "description"),
metadata=normed_metadata,
)
[docs]@whitelist_for_serdes(
storage_field_names={"metadata": "metadata_entries"},
field_serializers={"metadata": MetadataFieldSerializer},
)
class TypeCheck(
NamedTuple(
"_TypeCheck",
[
("success", PublicAttr[bool]),
("description", PublicAttr[Optional[str]]),
("metadata", PublicAttr[Mapping[str, MetadataValue]]),
],
)
):
"""Event corresponding to a successful typecheck.
Events of this type should be returned by user-defined type checks when they need to encapsulate
additional metadata about a type check's success or failure. (i.e., when using
:py:func:`as_dagster_type`, :py:func:`@usable_as_dagster_type <dagster_type>`, or the underlying
:py:func:`PythonObjectDagsterType` API.)
Op compute functions should generally avoid yielding events of this type to avoid confusion.
Args:
success (bool): ``True`` if the type check succeeded, ``False`` otherwise.
description (Optional[str]): A human-readable description of the type check.
metadata (Optional[Dict[str, RawMetadataValue]]):
Arbitrary metadata about the failure. Keys are displayed string labels, and values are
one of the following: string, float, int, JSON-serializable dict, JSON-serializable
list, and one of the data classes returned by a MetadataValue static method.
"""
def __new__(
cls,
success: bool,
description: Optional[str] = None,
metadata: Optional[Mapping[str, RawMetadataValue]] = None,
):
normed_metadata = normalize_metadata(
check.opt_mapping_param(metadata, "metadata", key_type=str),
)
return super(TypeCheck, cls).__new__(
cls,
success=check.bool_param(success, "success"),
description=check.opt_str_param(description, "description"),
metadata=normed_metadata,
)
[docs]class Failure(Exception):
"""Event indicating op failure.
Raise events of this type from within op compute functions or custom type checks in order to
indicate an unrecoverable failure in user code to the Dagster machinery and return
structured metadata about the failure.
Args:
description (Optional[str]): A human-readable description of the failure.
metadata (Optional[Dict[str, RawMetadataValue]]):
Arbitrary metadata about the failure. Keys are displayed string labels, and values are
one of the following: string, float, int, JSON-serializable dict, JSON-serializable
list, and one of the data classes returned by a MetadataValue static method.
allow_retries (Optional[bool]):
Whether this Failure should respect the retry policy or bypass it and immediately fail.
Defaults to True, respecting the retry policy and allowing retries.
"""
def __init__(
self,
description: Optional[str] = None,
metadata: Optional[Mapping[str, RawMetadataValue]] = None,
allow_retries: Optional[bool] = None,
):
super(Failure, self).__init__(description)
self.description = check.opt_str_param(description, "description")
self.metadata = normalize_metadata(
check.opt_mapping_param(metadata, "metadata", key_type=str),
)
self.allow_retries = check.opt_bool_param(allow_retries, "allow_retries", True)
[docs]class RetryRequested(Exception):
"""An exception to raise from an op to indicate that it should be retried.
Args:
max_retries (Optional[int]):
The max number of retries this step should attempt before failing
seconds_to_wait (Optional[Union[float,int]]):
Seconds to wait before restarting the step after putting the step in
to the up_for_retry state
Example:
.. code-block:: python
@op
def flakes():
try:
flakey_operation()
except Exception as e:
raise RetryRequested(max_retries=3) from e
"""
def __init__(
self, max_retries: Optional[int] = 1, seconds_to_wait: Optional[Union[float, int]] = None
):
super(RetryRequested, self).__init__()
self.max_retries = check.int_param(max_retries, "max_retries")
self.seconds_to_wait = check.opt_numeric_param(seconds_to_wait, "seconds_to_wait")
class ObjectStoreOperationType(Enum):
SET_OBJECT = "SET_OBJECT"
GET_OBJECT = "GET_OBJECT"
RM_OBJECT = "RM_OBJECT"
CP_OBJECT = "CP_OBJECT"
class ObjectStoreOperation(
NamedTuple(
"_ObjectStoreOperation",
[
("op", ObjectStoreOperationType),
("key", str),
("dest_key", Optional[str]),
("obj", Any),
("serialization_strategy_name", Optional[str]),
("object_store_name", Optional[str]),
("value_name", Optional[str]),
("version", Optional[str]),
("mapping_key", Optional[str]),
],
)
):
"""This event is used internally by Dagster machinery when values are written to and read from
an ObjectStore.
Users should not import this class or yield events of this type from user code.
Args:
op (ObjectStoreOperationType): The type of the operation on the object store.
key (str): The key of the object on which the operation was performed.
dest_key (Optional[str]): The destination key, if any, to which the object was copied.
obj (Any): The object, if any, retrieved by the operation.
serialization_strategy_name (Optional[str]): The name of the serialization strategy, if any,
employed by the operation
object_store_name (Optional[str]): The name of the object store that performed the
operation.
value_name (Optional[str]): The name of the input/output
version (Optional[str]): (Experimental) The version of the stored data.
mapping_key (Optional[str]): The mapping key when a dynamic output is used.
"""
def __new__(
cls,
op: ObjectStoreOperationType,
key: str,
dest_key: Optional[str] = None,
obj: Any = None,
serialization_strategy_name: Optional[str] = None,
object_store_name: Optional[str] = None,
value_name: Optional[str] = None,
version: Optional[str] = None,
mapping_key: Optional[str] = None,
):
return super(ObjectStoreOperation, cls).__new__(
cls,
op=op,
key=check.str_param(key, "key"),
dest_key=check.opt_str_param(dest_key, "dest_key"),
obj=obj,
serialization_strategy_name=check.opt_str_param(
serialization_strategy_name, "serialization_strategy_name"
),
object_store_name=check.opt_str_param(object_store_name, "object_store_name"),
value_name=check.opt_str_param(value_name, "value_name"),
version=check.opt_str_param(version, "version"),
mapping_key=check.opt_str_param(mapping_key, "mapping_key"),
)
@classmethod
def serializable(cls, inst, **kwargs):
return cls(
**dict(
{
"op": inst.op.value,
"key": inst.key,
"dest_key": inst.dest_key,
"obj": None,
"serialization_strategy_name": inst.serialization_strategy_name,
"object_store_name": inst.object_store_name,
"value_name": inst.value_name,
"version": inst.version,
},
**kwargs,
)
)
class HookExecutionResult(
NamedTuple("_HookExecutionResult", [("hook_name", str), ("is_skipped", bool)])
):
"""This event is used internally to indicate the execution result of a hook, e.g. whether the
user-defined hook function is skipped.
Args:
hook_name (str): The name of the hook.
is_skipped (bool): ``False`` if the hook_fn is executed, ``True`` otheriwse.
"""
def __new__(cls, hook_name: str, is_skipped: Optional[bool] = None):
return super(HookExecutionResult, cls).__new__(
cls,
hook_name=check.str_param(hook_name, "hook_name"),
is_skipped=cast(bool, check.opt_bool_param(is_skipped, "is_skipped", default=False)),
)
UserEvent = Union[AssetMaterialization, AssetObservation, ExpectationResult]