def test_cli_validate_exec(
trace_gdf,
assume_error: rules.ValidationResults,
pandera_valid: bool,
database,
tmp_path_factory,
metadata_json: Path,
):
"""
Test tracerepo validate command with a set up of invalidated data.
"""
assert metadata_json.exists() and metadata_json.is_file()
area_gdf: gpd.GeoDataFrame = tests.kb11_area
tmp_path = tmp_path_factory.mktemp(basename="test_cli_validate_exec", numbered=True)
assert len(list(tmp_path.glob("*"))) == 0
# Make default directories
repo.scaffold(tmp_path)
organizer = tests.set_up_repo_with_invalids_organized(
database=database,
trace_gdf=trace_gdf,
area_gdf=area_gdf,
tracerepository_path=tmp_path,
)
database_csv_path: Path = tmp_path / rules.DATABASE_CSV
if database_csv_path.exists():
database_csv_path.unlink()
repo.write_database_csv(path=database_csv_path, database=organizer.database)
# Test if all column headers are in csv
database_text = database_csv_path.read_text()
database_first_line = database_text.splitlines()[0]
for column_enum in rules.ColumnNames:
assert column_enum.value in database_first_line.split(",")
result = runner.invoke(
app=app,
args=[
"validate",
"--report",
f"--metadata-json={metadata_json}",
f"--tracerepository-path={tmp_path}",
],
)
reports_path = Path(tmp_path) / Path(rules.PathNames.REPORTS.value)
if not pandera_valid:
assert reports_path.exists()
assert reports_path.is_dir()
assert len(list(reports_path.glob("*.html"))) > 0
# TODO: Inconsistent results here
if not "html" in result.output:
warn(f"Unknown error, html not found in result.output.")
tests.click_error_print(result)
# TODO: Inconsistent results here
if assume_error.value not in database_csv_path.read_text():
warn(f"Expected {assume_error.value} to be in {database_csv_path}.")
def organizer_unorganized(tmp_path):
"""
Fix up an Organizer with files in unorganized.
"""
df = setup_df_with_traces_and_area(
repo.scaffold(tracerepository_path=tmp_path),
tracerepository_path=tmp_path)
repo.scaffold(tmp_path)
yield Organizer(database=df, tracerepository_path=tmp_path)
def df_with_traces_and_area(tmp_path):
"""
Populate scaffold and database with a single traces+area row.
traces and area file will be in unorganized.
"""
df = setup_df_with_traces_and_area(repo.scaffold(tmp_path),
tracerepository_path=tmp_path)
yield df
def init(
tracerepository_path: Path = TRACEREPOSITORY_PATH_OPTION,
database_name: str = DATABASE_OPTION,
):
"""
Initialize tracerepo in current directory.
"""
df = repo.scaffold(tracerepository_path=tracerepository_path)
repo.write_database_csv(path=tracerepository_path / database_name,
database=df)
def test_validate_invalids_with_full_setup(
database: pd.DataFrame,
tmp_path_factory: TempPathFactory,
trace_gdf: gpd.GeoDataFrame,
assume_error: rules.ValidationResults,
):
"""
Test validate_invalids.
"""
area_gdf: gpd.GeoDataFrame = tests.kb11_area
tmp_path = tmp_path_factory.mktemp(basename="test_validate_invalids", numbered=True)
repo.scaffold(tmp_path)
organizer = tests.set_up_repo_with_invalids_organized(
database=database,
trace_gdf=trace_gdf,
area_gdf=area_gdf,
tracerepository_path=tmp_path,
)
# Query for invalid traces
invalids = organizer.query(
validity=[rules.ValidationResults.INVALID],
)
for invalid in invalids:
assert invalid.traces_path.exists()
assert invalid.area_path.exists()
update_tuples = spatial.validate_invalids(invalids)
assert isinstance(update_tuples, list)
assert all(isinstance(val, utils.UpdateTuple) for val in update_tuples)
for update_tuple in update_tuples:
assert (
update_tuple.update_values[rules.ColumnNames.VALIDITY] == assume_error.value
)
def test_cli_organize(database, trace_gdf, other_args, tmp_path_factory):
"""
Test cli_organize click entrypoint.
"""
area_gdf: gpd.GeoDataFrame = tests.kb11_area
tmp_path = tmp_path_factory.mktemp(basename="test_cli_organize", numbered=True)
other_args.append(f"--tracerepository-path={tmp_path}")
args = ["organize"] + other_args
# make default directories
repo.scaffold(tmp_path)
organizer = tests.set_up_repo_with_invalids_organized(
database=database,
trace_gdf=trace_gdf,
area_gdf=area_gdf,
organized=False,
tracerepository_path=tmp_path,
)
repo.write_database_csv(
path=tmp_path / rules.DATABASE_CSV, database=organizer.database
)
result = runner.invoke(app=app, args=args)
if "--simulate" not in args:
organizer.check()
if "--report" in args:
assert len(result.stdout) > 0
elif "--no-report" in args:
assert len(result.stdout) == 0
tests.click_error_print(result)
def organizer_organized(tmp_path_factory: TempPathFactory):
"""
Fix up an Organizer with files organized.
"""
tmp_path = tmp_path_factory.mktemp(basename="organizer_organized")
df = setup_df_with_traces_and_area(
repo.scaffold(tracerepository_path=tmp_path),
tracerepository_path=tmp_path)
# with tests.setup_scaffold_context(tmp_path) as df:
# df = setup_df_with_traces_and_area(df=df)
organizer = Organizer(database=df, tracerepository_path=tmp_path)
organizer.organize(simulate=False)
organizer.check()
yield organizer