113 lines
3 KiB
Python
113 lines
3 KiB
Python
from __future__ import annotations
|
|
|
|
import zipfile
|
|
import json
|
|
|
|
import polars as pl
|
|
|
|
from pipeline.validate_outputs import main
|
|
|
|
|
|
def write_boundary(path, postcodes):
|
|
units = path / "units"
|
|
units.mkdir(parents=True)
|
|
features = [
|
|
{"type": "Feature", "properties": {"postcodes": postcode}, "geometry": None}
|
|
for postcode in postcodes
|
|
]
|
|
(units / "AA1.geojson").write_text(
|
|
json.dumps({"type": "FeatureCollection", "features": features})
|
|
)
|
|
|
|
|
|
def test_validates_parquet_file_and_zip(tmp_path, monkeypatch):
|
|
parquet_path = tmp_path / "data.parquet"
|
|
file_path = tmp_path / "plain.txt"
|
|
zip_path = tmp_path / "archive.zip"
|
|
|
|
pl.DataFrame({"value": [1]}).write_parquet(parquet_path)
|
|
file_path.write_text("ok\n")
|
|
with zipfile.ZipFile(zip_path, "w") as archive:
|
|
archive.writestr("data.txt", "ok\n")
|
|
|
|
monkeypatch.setattr(
|
|
"sys.argv",
|
|
[
|
|
"validate_outputs",
|
|
"--parquet",
|
|
str(parquet_path),
|
|
"--file",
|
|
str(file_path),
|
|
"--zip",
|
|
str(zip_path),
|
|
"--glob",
|
|
f"{tmp_path}::*.txt",
|
|
"--zip-glob",
|
|
f"{tmp_path}::*.zip",
|
|
],
|
|
)
|
|
|
|
assert main() == 0
|
|
|
|
|
|
def test_rejects_missing_and_empty_outputs(tmp_path, monkeypatch, capsys):
|
|
empty_path = tmp_path / "empty.txt"
|
|
empty_path.touch()
|
|
|
|
monkeypatch.setattr(
|
|
"sys.argv",
|
|
[
|
|
"validate_outputs",
|
|
"--file",
|
|
str(empty_path),
|
|
"--parquet",
|
|
str(tmp_path / "missing.parquet"),
|
|
"--glob",
|
|
f"{tmp_path}::*.csv",
|
|
],
|
|
)
|
|
|
|
assert main() == 1
|
|
stderr = capsys.readouterr().err
|
|
assert "empty file" in stderr
|
|
assert "missing" in stderr
|
|
assert "no files matched" in stderr
|
|
|
|
|
|
def test_validates_postcode_boundary_matches(tmp_path, monkeypatch):
|
|
postcodes_path = tmp_path / "postcodes.parquet"
|
|
boundaries_path = tmp_path / "postcode_boundaries"
|
|
pl.DataFrame({"postcode": ["AA1 1AA", "AA1 1AB"]}).write_parquet(postcodes_path)
|
|
write_boundary(boundaries_path, ["AA1 1AA", "AA1 1AB"])
|
|
|
|
monkeypatch.setattr(
|
|
"sys.argv",
|
|
[
|
|
"validate_outputs",
|
|
"--postcode-boundary-match",
|
|
f"{postcodes_path}::{boundaries_path}",
|
|
],
|
|
)
|
|
|
|
assert main() == 0
|
|
|
|
|
|
def test_rejects_postcode_boundary_mismatch(tmp_path, monkeypatch, capsys):
|
|
postcodes_path = tmp_path / "postcodes.parquet"
|
|
boundaries_path = tmp_path / "postcode_boundaries"
|
|
pl.DataFrame({"postcode": ["AA1 1AA", "AA1 1AB"]}).write_parquet(postcodes_path)
|
|
write_boundary(boundaries_path, ["AA1 1AA", "AA1 1AC"])
|
|
|
|
monkeypatch.setattr(
|
|
"sys.argv",
|
|
[
|
|
"validate_outputs",
|
|
"--postcode-boundary-match",
|
|
f"{postcodes_path}::{boundaries_path}",
|
|
],
|
|
)
|
|
|
|
assert main() == 1
|
|
stderr = capsys.readouterr().err
|
|
assert "missing boundaries" in stderr
|
|
assert "boundary postcodes are absent" in stderr
|