Support NamedTuples for named arguments in parametrizations. #224
Description
Workarounds
typing.NamedTuple or collections.namedtuple can be used to achieve named arguments per iteration of the parametrization.
from pathlib import Path
from typing import NamedTuple
class Task(NamedTuple):
depends_on: Path
produces: Path
@pytask.mark.parametrize("depends_on, produces", [
Task(depends_on="first_dataset.pkl", produces="first_plot.png"),
Task(depends_on="second_dataset.pkl", produces="second_plot.png"),
])
def task_plot_data(depends_on, produces):
df = pd.read_pickle(depends_on)
ax = df.plot()
plt.savefig(produces)- Document example usage in the documentation.
- Implement a check for whether the arguments of the namedtuple match the signature of the parametrization.
Originally posted by @tobiasraabe in #206 (comment)
Activity
tobiasraabe commentedon Mar 3, 2022
This will not be pursued. #229 implements a new approach to parametrizations which will eventually supersede pytest's approach.
We keep this issue open for documentation. Probably until the pytest approach is deprecated.