Rename convert_vcf to convert

Author: jeromekelleher

bio2zarr/cli.py

@@ -85,7 +85,7 @@ def encode(if_path, zarr_path, verbose, schema, worker_processes):
 @worker_processes
 def convert_vcf(vcfs, out_path, verbose, worker_processes):
     setup_logging(verbose)
-    vcf.convert_vcf(
+    vcf.convert(
         vcfs, out_path, show_progress=True, worker_processes=worker_processes
     )
 

bio2zarr/vcf.py

@@ -1344,7 +1344,7 @@ def encode(
     )
 
 
-def convert_vcf(
+def convert(
     vcfs,
     out_path,
     *,

tests/test_cli.py

@@ -64,3 +64,19 @@ def test_encode(self):
                 worker_processes=1,
                 show_progress=True,
             )
+
+    def test_convert(self):
+        runner = ct.CliRunner(mix_stderr=False)
+        with mock.patch("bio2zarr.vcf.convert") as mocked:
+            result = runner.invoke(
+                cli.vcf2zarr, ["convert", "vcf_path", "zarr_path"], catch_exceptions=False
+            )
+            assert result.exit_code == 0
+            assert len(result.stdout) == 0
+            assert len(result.stderr) == 0
+            mocked.assert_called_once_with(
+                ("vcf_path",),
+                "zarr_path",
+                worker_processes=1,
+                show_progress=True,
+            )

tests/test_simulated_data.py

@@ -28,7 +28,7 @@ def test_ploidy(self, ploidy, tmp_path):
         # This also compresses the input file
         pysam.tabix_index(str(vcf_file), preset="vcf")
         out = tmp_path / "example.vcf.zarr"
-        vcf.convert_vcf([tmp_path / "sim.vcf.gz"], out)
+        vcf.convert([tmp_path / "sim.vcf.gz"], out)
         ds = sg.load_dataset(out)
         assert ds.sizes["ploidy"] == ploidy
         assert ds.sizes["variants"] == ts.num_sites

tests/test_vcf_examples.py

@@ -13,7 +13,7 @@ class TestSmallExample:
     @pytest.fixture(scope="class")
     def ds(self, tmp_path_factory):
         out = tmp_path_factory.mktemp("data") / "example.vcf.zarr"
-        vcf.convert_vcf([self.data_path], out)
+        vcf.convert([self.data_path], out)
         return sg.load_dataset(out)
 
     def test_filters(self, ds):
@@ -224,7 +224,7 @@ def test_call_HQ(self, ds):
     def test_no_genotypes(self, ds, tmp_path):
         path = "tests/data/vcf/sample_no_genotypes.vcf.gz"
         out = tmp_path / "example.vcf.zarr"
-        vcf.convert_vcf([path], out)
+        vcf.convert([path], out)
         ds2 = sg.load_dataset(out)
         assert len(ds2["sample_id"]) == 0
         for col in ds:
@@ -244,7 +244,7 @@ def test_chunk_size(
         self, ds, tmp_path, chunk_length, chunk_width, y_chunks, x_chunks
     ):
         out = tmp_path / "example.vcf.zarr"
-        vcf.convert_vcf(
+        vcf.convert(
             [self.data_path], out, chunk_length=chunk_length, chunk_width=chunk_width
         )
         ds2 = sg.load_dataset(out)
@@ -277,7 +277,7 @@ def test_chunk_size(
     @pytest.mark.parametrize("worker_processes", [0, 1, 2])
     def test_worker_processes(self, ds, tmp_path, worker_processes):
         out = tmp_path / "example.vcf.zarr"
-        vcf.convert_vcf(
+        vcf.convert(
             [self.data_path], out, chunk_length=3, worker_processes=worker_processes
         )
         ds2 = sg.load_dataset(out)
@@ -306,7 +306,7 @@ class Test1000G2020Example:
     @pytest.fixture(scope="class")
     def ds(self, tmp_path_factory):
         out = tmp_path_factory.mktemp("data") / "example.vcf.zarr"
-        vcf.convert_vcf([self.data_path], out, worker_processes=0)
+        vcf.convert([self.data_path], out, worker_processes=0)
         return sg.load_dataset(out)
 
     def test_position(self, ds):
@@ -418,7 +418,7 @@ class Test1000G2020AnnotationsExample:
     def ds(self, tmp_path_factory):
         out = tmp_path_factory.mktemp("data") / "example.zarr"
         # TODO capture warnings from htslib here
-        vcf.convert_vcf([self.data_path], out, worker_processes=0)
+        vcf.convert([self.data_path], out, worker_processes=0)
         return sg.load_dataset(out)
 
     def test_position(self, ds):
@@ -650,7 +650,7 @@ class TestGeneratedFieldsExample:
     @pytest.fixture(scope="class")
     def ds(self, tmp_path_factory):
         out = tmp_path_factory.mktemp("data") / "vcf.zarr"
-        vcf.convert_vcf([self.data_path], out)
+        vcf.convert([self.data_path], out)
         return sg.load_dataset(out)
 
     def test_info_string1(self, ds):
@@ -700,5 +700,5 @@ def test_info_string2(self, ds):
 def test_by_validating(name, tmp_path):
     path = f"tests/data/vcf/{name}"
     out = tmp_path / "test.zarr"
-    vcf.convert_vcf([path], out, worker_processes=0)
+    vcf.convert([path], out, worker_processes=0)
     vcf.validate(path, out)