Rename IndexedVcf to VcfFile

Author: jeromekelleher

bio2zarr/cli.py

@@ -624,8 +624,8 @@ def vcfpartition(vcfs, verbose, num_partitions, partition_size):
         num_parts_per_path = max(1, num_partitions // len(vcfs))
 
     for vcf_path in vcfs:
-        indexed_vcf = vcf_utils.IndexedVcf(vcf_path)
-        regions = indexed_vcf.partition_into_regions(
+        vcf_file = vcf_utils.VcfFile(vcf_path)
+        regions = vcf_file.partition_into_regions(
             num_parts=num_parts_per_path, target_part_size=partition_size
         )
         for region in regions:

bio2zarr/vcf2zarr/icf.py

@@ -228,8 +228,8 @@ def make_field_def(name, vcf_type, vcf_number):
 
 
 def scan_vcf(path, target_num_partitions):
-    with vcf_utils.IndexedVcf(path) as indexed_vcf:
-        vcf = indexed_vcf.vcf
+    with vcf_utils.VcfFile(path) as vcf_file:
+        vcf = vcf_file.vcf
         filters = []
         pass_index = -1
         for h in vcf.header_iter():
@@ -270,10 +270,10 @@ def scan_vcf(path, target_num_partitions):
             filters=filters,
             fields=fields,
             partitions=[],
-            num_records=sum(indexed_vcf.contig_record_counts().values()),
+            num_records=sum(vcf_file.contig_record_counts().values()),
         )
 
-        regions = indexed_vcf.partition_into_regions(num_parts=target_num_partitions)
+        regions = vcf_file.partition_into_regions(num_parts=target_num_partitions)
         for region in regions:
             metadata.partitions.append(
                 VcfPartition(
@@ -1093,9 +1093,9 @@ def process_partition(self, partition_index):
             self.path,
             partition_index,
         ) as tcw:
-            with vcf_utils.IndexedVcf(partition.vcf_path) as ivcf:
+            with vcf_utils.VcfFile(partition.vcf_path) as vcf:
                 num_records = 0
-                for variant in ivcf.variants(partition.region):
+                for variant in vcf.variants(partition.region):
                     num_records += 1
                     last_position = variant.POS
                     tcw.append("CHROM", variant.CHROM)

bio2zarr/vcf_utils.py

@@ -396,7 +396,7 @@ class VcfIndexType(Enum):
     TABIX = ".tbi"
 
 
-class IndexedVcf(contextlib.AbstractContextManager):
+class VcfFile(contextlib.AbstractContextManager):
     def __init__(self, vcf_path, index_path=None):
         self.vcf = None
         self.file_type = None
@@ -471,7 +471,7 @@ def __exit__(self, exc_type, exc_val, exc_tb):
 
     def contig_record_counts(self):
         if self.index is None:
-            return {self.sequence_names[0]: np.inf}
+            return {self.sequence_names[0]: RECORD_COUNT_UNKNOWN}
         d = dict(zip(self.sequence_names, self.index.record_counts))
         if self.file_type == VcfFileType.BCF:
             d = {k: v for k, v in d.items() if v > 0}

tests/test_vcf_utils.py

@@ -14,24 +14,24 @@ def assert_part_counts_non_zero(part_counts, index_file):
     assert np.all(part_counts > 0)
 
 
-class TestIndexedVcf:
+class TestVcfFile:
     def get_instance(self, index_file):
         vcf_path = data_path / (".".join(list(index_file.split("."))[:-1]))
-        return vcf_utils.IndexedVcf(vcf_path, data_path / index_file)
+        return vcf_utils.VcfFile(vcf_path, data_path / index_file)
 
     def test_context_manager_success(self):
         # Nominal case
-        with vcf_utils.IndexedVcf(data_path / "sample.bcf") as iv:
+        with vcf_utils.VcfFile(data_path / "sample.bcf") as iv:
             assert iv.vcf is not None
         assert iv.vcf is None
 
     def test_context_manager_error(self):
         with pytest.raises(FileNotFoundError, match="no-such-file"):
-            with vcf_utils.IndexedVcf(data_path / "no-such-file.bcf"):
+            with vcf_utils.VcfFile(data_path / "no-such-file.bcf"):
                 pass
 
     def test_indels_filtered(self):
-        with vcf_utils.IndexedVcf(data_path / "chr_m_indels.vcf.gz") as vfile:
+        with vcf_utils.VcfFile(data_path / "chr_m_indels.vcf.gz") as vfile:
             # Hand-picked example that results in filtering
             region = vcf_utils.Region("chrM", 300, 314)
             pos = [var.POS for var in vfile.variants(region)]
@@ -173,7 +173,7 @@ def test_partition_into_n_parts_unindexed(
     ):
         copy_path = tmp_path / vcf_file
         shutil.copyfile(data_path / vcf_file, copy_path)
-        indexed_vcf = vcf_utils.IndexedVcf(copy_path)
+        indexed_vcf = vcf_utils.VcfFile(copy_path)
         regions = list(indexed_vcf.partition_into_regions(num_parts=num_parts))
         assert len(regions) == 1
         part_variant_counts = np.array(
@@ -238,12 +238,12 @@ def test_partition_invalid_arguments(self):
     @pytest.mark.parametrize("path", ["y", data_path / "xxx", "/x/y.csi"])
     def test_missing_index_file(self, path):
         with pytest.raises(FileNotFoundError, match="Specified index path"):
-            vcf_utils.IndexedVcf(data_path / "sample.vcf.gz", path)
+            vcf_utils.VcfFile(data_path / "sample.vcf.gz", path)
 
     def test_bad_index_format(self):
         vcf_file = data_path / "sample.vcf.gz"
         with pytest.raises(ValueError, match="Only .tbi or .csi indexes"):
-            vcf_utils.IndexedVcf(vcf_file, vcf_file)
+            vcf_utils.VcfFile(vcf_file, vcf_file)
 
     @pytest.mark.parametrize(
         "filename",
@@ -256,11 +256,11 @@ def test_bad_index_format(self):
         ],
     )
     def test_unindexed_single_contig(self, tmp_path, filename):
-        f1 = vcf_utils.IndexedVcf(data_path / filename)
+        f1 = vcf_utils.VcfFile(data_path / filename)
         assert f1.index is not None
         copy_path = tmp_path / filename
         shutil.copyfile(data_path / filename, copy_path)
-        f2 = vcf_utils.IndexedVcf(copy_path)
+        f2 = vcf_utils.VcfFile(copy_path)
         assert f2.index is None
         crc1 = f1.contig_record_counts()
         assert len(crc1) == 1
@@ -280,7 +280,7 @@ def test_unindexed_single_contig(self, tmp_path, filename):
     def test_unindexed_multi_contig(self, tmp_path, filename):
         copy_path = tmp_path / filename
         shutil.copyfile(data_path / filename, copy_path)
-        f = vcf_utils.IndexedVcf(copy_path)
+        f = vcf_utils.VcfFile(copy_path)
         with pytest.raises(ValueError, match="Multi-contig VCFs must be indexed"):
             list(f.variants())