Create a repository from the decorated function.
The decorated function should take no arguments and its return value should one of:
1. List[Union[JobDefinition, ScheduleDefinition, SensorDefinition]].
Use this form when you have no need to lazy load pipelines or other definitions. This is the
typical use case.
A dict of the form:
{
'jobs': Dict[str, Callable[[], JobDefinition]],
'schedules': Dict[str, Callable[[], ScheduleDefinition]]
'sensors': Dict[str, Callable[[], SensorDefinition]]
}
This form is intended to allow definitions to be created lazily when accessed by name, which can be helpful for performance when there are many definitions in a repository, or when constructing the definitions is costly.
3. A RepositoryData. Return this object if you need fine-grained
control over the construction and indexing of definitions within the repository, e.g., to
create definitions dynamically from .yaml files in a directory.
name (Optional[str]) – The name of the repository. Defaults to the name of the decorated function.
description (Optional[str]) – A string description of the repository.
top_level_resources (Optional[Mapping[str, ResourceDefinition]]) – A dict of top-level resource keys to defintions, for resources which should be displayed in the UI.
Example
######################################################################
# A simple repository using the first form of the decorated function
######################################################################
@op(config_schema={n: Field(Int)})
def return_n(context):
return context.op_config['n']
@job
def simple_job():
return_n()
@job
def some_job():
...
@sensor(job=some_job)
def some_sensor():
if foo():
yield RunRequest(
run_key= ...,
run_config={
'ops': {'return_n': {'config': {'n': bar()}}}
}
)
@job
def my_job():
...
my_schedule = ScheduleDefinition(cron_schedule="0 0 * * *", job=my_job)
@repository
def simple_repository():
return [simple_job, some_sensor, my_schedule]
######################################################################
# A lazy-loaded repository
######################################################################
def make_expensive_job():
@job
def expensive_job():
for i in range(10000):
return_n.alias(f'return_n_{i}')()
return expensive_job
def make_expensive_schedule():
@job
def other_expensive_job():
for i in range(11000):
return_n.alias(f'my_return_n_{i}')()
return ScheduleDefinition(cron_schedule="0 0 * * *", job=other_expensive_job)
@repository
def lazy_loaded_repository():
return {
'jobs': {'expensive_job': make_expensive_job},
'schedules': {'expensive_schedule': make_expensive_schedule}
}
######################################################################
# A complex repository that lazily constructs jobs from a directory
# of files in a bespoke YAML format
######################################################################
class ComplexRepositoryData(RepositoryData):
def __init__(self, yaml_directory):
self._yaml_directory = yaml_directory
def get_all_pipelines(self):
return [
self._construct_job_def_from_yaml_file(
self._yaml_file_for_job_name(file_name)
)
for file_name in os.listdir(self._yaml_directory)
]
...
@repository
def complex_repository():
return ComplexRepositoryData('some_directory')
Define a repository that contains a group of definitions.
Users should typically not create objects of this class directly. Instead, use the
@repository() decorator.
name (str) – The name of the repository.
repository_data (RepositoryData) – Contains the definitions making up the repository.
description (Optional[str]) – A string description of the repository.
Return all jobs in the repository as a list.
Note that this will construct any job in the lazily evaluated dictionary that has not yet been constructed.
All jobs in the repository.
List[JobDefinition]
Returns an object that can load the contents of assets as Python objects.
Invokes load_input on the IOManager associated with the assets. Avoids
spinning up resources separately for each asset.
Usage:
with my_repo.get_asset_value_loader() as loader:
asset1 = loader.load_asset_value("asset1")
asset2 = loader.load_asset_value("asset2")
Get a job by name.
If this job is present in the lazily evaluated dictionary passed to the constructor, but has not yet been constructed, only this job is constructed, and will be cached for future calls.
name (str) – Name of the job to retrieve.
The job definition corresponding to the given name.
Check if a job with a given name is present in the repository.
name (str) – The name of the job.
bool
Names of all jobs in the repository.
List[str]
Load the contents of an asset as a Python object.
Invokes load_input on the IOManager associated with the asset.
If you want to load the values of multiple assets, it’s more efficient to use
get_asset_value_loader(), which avoids spinning up
resources separately for each asset.
asset_key (Union[AssetKey, Sequence[str], str]) – The key of the asset to load.
python_type (Optional[Type]) – The python type to load the asset as. This is what will be returned inside load_input by context.dagster_type.typing_type.
partition_key (Optional[str]) – The partition of the asset to load.
resource_config (Optional[Any]) – A dictionary of resource configurations to be passed
to the IOManager.
The contents of an asset as a Python object.
Users should usually rely on the @repository decorator to create new
repositories, which will in turn call the static constructors on this class. However, users may
subclass RepositoryData for fine-grained control over access to and lazy creation
of repository members.
Return all jobs in the repository as a list.
All jobs in the repository.
List[JobDefinition]
Return all schedules in the repository as a list.
All pipelines in the repository.
List[ScheduleDefinition]
Get a job by name.
job_name (str) – Name of the job to retrieve.
The job definition corresponding to the given name.