def test_build_url():
assert build_url("http://host/path", "subpath") == "http://host/path/subpath"
assert build_url("http://host/path/", "subpath") == "http://host/path/subpath"
assert (
build_url("http://host/path?a=b", "subpath") == "http://host/path/subpath?a=b"
)
assert (
build_url("http://host/path/?a=b", "subpath") == "http://host/path/subpath?a=b"
)
assert build_url("http://host/path#a", "subpath") == "http://host/path/subpath#a"
assert build_url("s3://host/path", "subpath") == "s3://host/path/subpath"
assert build_url("relative_path/path", "subpath") == "relative_path/path/subpath"
assert build_url("/absolute_path/path", "subpath") == "/absolute_path/path/subpath"
assert (
build_url("http://host/a%20path", "subpath") == "http://host/a%20path/subpath"
)
assert build_url("http://host/a path", "subpath") == "http://host/a%20path/subpath"
f"For multiple inputs, multiple input regions must be a sequence of sequence of strings: {regions}"
)
input_regions = regions
assert len(inputs) == len(input_regions)
tasks = []
parts = []
for i, input in enumerate(inputs):
filename = url_filename(str(input))
input_region_list = input_regions[i]
if input_region_list is None:
# single partition case: make a list so the loop below works
input_region_list = [None] # type: ignore
for r, region in enumerate(input_region_list):
part_url = build_url(str(output), f"{filename}/part-{r}.zarr")
output_part = fsspec.get_mapper(part_url, **output_storage_options)
parts.append(part_url)
task = dask.delayed(vcf_to_zarr_sequential)(
input,
output=output_part,
region=region,
chunk_length=chunk_length,
chunk_width=chunk_width,
ploidy=ploidy,
mixed_ploidy=mixed_ploidy,
truncate_calls=truncate_calls,
)
tasks.append(task)
dask.compute(*tasks)
return parts