Rename "to_zarr" to "encode"

Author: jeromekelleher

bio2zarr/cli.py

@@ -65,14 +65,14 @@ def mkschema(if_path):
 @click.argument("if_path", type=click.Path())
 @click.argument("zarr_path", type=click.Path())
 @verbose
-@click.option("-s", "--conversion-spec", default=None)
+@click.option("-s", "--schema", default=None)
 @worker_processes
-def to_zarr(if_path, zarr_path, verbose, conversion_spec, worker_processes):
+def encode(if_path, zarr_path, verbose, schema, worker_processes):
     setup_logging(verbose)
-    vcf.to_zarr(
+    vcf.encode(
         if_path,
         zarr_path,
-        conversion_spec,
+        schema,
         worker_processes=worker_processes,
         show_progress=True,
     )
@@ -105,7 +105,7 @@ def vcf2zarr():
 vcf2zarr.add_command(explode)
 vcf2zarr.add_command(inspect)
 vcf2zarr.add_command(mkschema)
-vcf2zarr.add_command(to_zarr)
+vcf2zarr.add_command(encode)
 vcf2zarr.add_command(convert_vcf)
 vcf2zarr.add_command(validate)
 

bio2zarr/vcf.py

@@ -1326,19 +1326,19 @@ def mkschema(if_path, out):
     out.write(spec.asjson())
 
 
-def to_zarr(
-    if_path, zarr_path, conversion_spec, worker_processes=1, show_progress=False
+def encode(
+    if_path, zarr_path, schema_path, worker_processes=1, show_progress=False
 ):
     pcvcf = PickleChunkedVcf.load(if_path)
-    if conversion_spec is None:
-        spec = ZarrConversionSpec.generate(pcvcf)
+    if schema_path is None:
+        schema = ZarrConversionSpec.generate(pcvcf)
     else:
-        with open(conversion_spec, "r") as f:
-            spec = ZarrConversionSpec.fromjson(f.read())
+        with open(schema_path, "r") as f:
+            schema = ZarrConversionSpec.fromjson(f.read())
     SgvcfZarr.convert(
         pcvcf,
         zarr_path,
-        conversion_spec=spec,
+        conversion_spec=schema,
         worker_processes=worker_processes,
         show_progress=show_progress,
     )

tests/test_cli.py

@@ -47,3 +47,20 @@ def test_mkschema(self):
             # the CliRunner
             # mocked.assert_called_once_with("path", stdout)
             mocked.assert_called_once()
+
+    def test_encode(self):
+        runner = ct.CliRunner(mix_stderr=False)
+        with mock.patch("bio2zarr.vcf.encode") as mocked:
+            result = runner.invoke(
+                cli.vcf2zarr, ["encode", "if_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(
+                "if_path",
+                "zarr_path",
+                None,
+                worker_processes=1,
+                show_progress=True,
+            )

tests/test_vcf_examples.py

@@ -295,7 +295,7 @@ def test_full_pipeline(self, ds, tmp_path, worker_processes):
         with open(schema, "w") as f:
             vcf.mkschema(exploded, f)
         out = tmp_path / "example.zarr"
-        vcf.to_zarr(exploded, out, schema, worker_processes=worker_processes)
+        vcf.encode(exploded, out, schema, worker_processes=worker_processes)
         ds2 = sg.load_dataset(out)
         xt.assert_equal(ds, ds2)