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_tbi(shared_datadir, vcf_file, is_path):
# Check record counts in tabix with actual count of VCF
vcf_path = path_for_test(shared_datadir, vcf_file, is_path)
tabix_path = get_tabix_path(vcf_path)
assert tabix_path is not None
tabix = read_tabix(tabix_path)
for i, contig in enumerate(tabix.sequence_names):
assert tabix.record_counts[i] == count_variants(vcf_path, contig)
def test_read_tabix__invalid_tbi(shared_datadir, file, is_path):
with pytest.raises(ValueError, match=r"File not in Tabix format."):
read_tabix(path_for_test(shared_datadir, file, is_path))