Feat: add execute_flux

This commit is contained in:
Bertrand Benjamin 2025-01-04 15:30:32 +01:00
parent 1446c166ca
commit d04bfe1d44
2 changed files with 53 additions and 15 deletions

View File

@ -1,5 +1,6 @@
from plesna.compute.consume_flux import consume_flux
from plesna.graph.graph_set import GraphSet from plesna.graph.graph_set import GraphSet
from plesna.models.flux import Flux from plesna.models.flux import Flux, FluxMetaData
from plesna.storage.repository.repository import Repository from plesna.storage.repository.repository import Repository
@ -41,3 +42,8 @@ class DataPlateform:
def flux(self, name: str) -> Flux: def flux(self, name: str) -> Flux:
return self._fluxes[name] return self._fluxes[name]
def execute_flux(self, name: str) -> FluxMetaData:
if name not in self._fluxes:
raise DataPlateformError("The flux {name} is not registered")
return consume_flux(self._fluxes[name])

View File

@ -54,21 +54,21 @@ def test_add_repository(
@pytest.fixture @pytest.fixture
def dataplatform( def foo_flux(repository: FSRepository) -> Flux:
repository: FSRepository, src = {"username": repository.table("raw", "username")}
) -> DataPlateform: targets = {"username": repository.table("bronze", "username")}
dp = DataPlateform()
dp.add_repository("test", repository)
return dp
def foo(sources, targets):
return {"who": "foo"}
def test_listing_content(dataplatform: DataPlateform): extra_kwrds = {}
assert dataplatform.repository("test").schemas() == ["raw", "bronze", "silver"]
assert dataplatform.repository("test").schema("raw").tables == [ flux = Flux(
"recovery", sources=src,
"username", targets=targets,
"salary", transformation=Transformation(function=foo, extra_kwrds=extra_kwrds),
] )
return flux
@pytest.fixture @pytest.fixture
@ -89,7 +89,10 @@ def copy_flux(repository: FSRepository) -> Flux:
return flux return flux
def test_add_flux(dataplatform: DataPlateform, copy_flux: Flux): def test_add_flux(repository: FSRepository, copy_flux: Flux):
dataplatform = DataPlateform()
dataplatform.add_repository("test", repository)
dataplatform.add_flux(name="copy_flux", flux=copy_flux) dataplatform.add_flux(name="copy_flux", flux=copy_flux)
assert dataplatform.fluxes == ["copy_flux"] assert dataplatform.fluxes == ["copy_flux"]
dataplatform.add_flux(name="copy_flux_bis", flux=copy_flux) dataplatform.add_flux(name="copy_flux_bis", flux=copy_flux)
@ -97,3 +100,32 @@ def test_add_flux(dataplatform: DataPlateform, copy_flux: Flux):
assert dataplatform.flux("copy_flux") == copy_flux assert dataplatform.flux("copy_flux") == copy_flux
assert dataplatform.flux("copy_flux_bis") == copy_flux assert dataplatform.flux("copy_flux_bis") == copy_flux
@pytest.fixture
def dataplatform(
repository: FSRepository,
foo_flux: Flux,
copy_flux: Flux,
) -> DataPlateform:
dp = DataPlateform()
dp.add_repository("test", repository)
dp.add_flux("foo", foo_flux)
dp.add_flux("copy", copy_flux)
return dp
def test_listing_content(dataplatform: DataPlateform):
assert dataplatform.repository("test").schemas() == ["raw", "bronze", "silver"]
assert dataplatform.repository("test").schema("raw").tables == [
"recovery",
"username",
"salary",
]
def test_execute_flux(dataplatform: DataPlateform):
meta = dataplatform.execute_flux("foo")
assert meta.data == {"who": "foo"}