def read_index(index_path: PathType,
storage_options: Optional[Dict[str, str]] = None) -> Any:
url = str(index_path)
if url.endswith(TABIX_EXTENSION):
return read_tabix(url, storage_options=storage_options)
elif url.endswith(CSI_EXTENSION):
return read_csi(url, storage_options=storage_options)
else:
raise ValueError("Only .tbi or .csi indexes are supported.")
def test_record_counts_csi(shared_datadir, vcf_file, is_path):
# Check record counts in csi with actual count of VCF
vcf_path = path_for_test(shared_datadir, vcf_file, is_path)
csi_path = get_csi_path(vcf_path)
assert csi_path is not None
csi = read_csi(csi_path)
for i, contig in enumerate(VCF(vcf_path).seqnames):
assert csi.record_counts[i] == count_variants(vcf_path, contig)
def test_read_csi__invalid_csi(shared_datadir, file, is_path):
with pytest.raises(ValueError, match=r"File not in CSI format."):
read_csi(path_for_test(shared_datadir, file, is_path))