Add single argument definition for VCFs, and declare path type

Author: jeromekelleher

bio2zarr/cli.py

@@ -18,6 +18,12 @@ def list_commands(self, ctx):
 
 
 # Common arguments/options
+vcfs = click.argument(
+    "vcfs", nargs=-1, required=True, type=click.Path(exists=True, dir_okay=False)
+)
+
+icf_path = click.argument("icf_path", type=click.Path())
+
 verbose = click.option("-v", "--verbose", count=True, help="Increase verbosity")
 
 version = click.version_option(version=f"{provenance.__version__}")
@@ -65,7 +71,7 @@ def setup_logging(verbosity):
 
 
 @click.command
-@click.argument("vcfs", nargs=-1, required=True)
+@vcfs
 @click.argument("zarr_path", type=click.Path())
 @verbose
 @worker_processes
@@ -85,8 +91,8 @@ def explode(vcfs, zarr_path, verbose, worker_processes, column_chunk_size):
 
 
 @click.command
-@click.argument("vcfs", nargs=-1, required=True)
-@click.argument("icf_path", type=click.Path())
+@vcfs
+@icf_path
 @click.argument("num_partitions", type=int)
 @column_chunk_size
 @verbose
@@ -95,7 +101,7 @@ def dexplode_init(
     vcfs, icf_path, num_partitions, column_chunk_size, verbose, worker_processes
 ):
     """
-    Initial step for parallel conversion of VCF(s) to intermediate columnar format
+    Initial step for distributed conversion of VCF(s) to intermediate columnar format
     over the requested number of paritions.
     """
     setup_logging(verbose)
@@ -111,12 +117,12 @@ def dexplode_init(
 
 
 @click.command
-@click.argument("icf_path", type=click.Path())
+@icf_path
 @click.argument("partition", type=int)
 @verbose
 def dexplode_partition(icf_path, partition, verbose):
     """
-    Convert a VCF partition into intermediate columnar format. Must be called *after*
+    Convert a VCF partition to intermediate columnar format. Must be called *after*
     the ICF path has been initialised with dexplode_init. Partition indexes must be
     from 0 (inclusive) to the number of paritions returned by dexplode_init (exclusive).
     """
@@ -129,26 +135,26 @@ def dexplode_partition(icf_path, partition, verbose):
 @verbose
 def dexplode_finalise(path, verbose):
     """
-    Final step for parallel conversion of VCF(s) to intermediate columnar format
+    Final step for distributed conversion of VCF(s) to intermediate columnar format.
     """
     setup_logging(verbose)
     vcf.explode_finalise(path)
 
 
 @click.command
-@click.argument("icf_path", type=click.Path())
+@click.argument("path", type=click.Path())
 @verbose
-def inspect(icf_path, verbose):
+def inspect(path, verbose):
     """
-    Inspect an intermediate format or Zarr path.
+    Inspect an intermediate columnar format or Zarr path.
     """
     setup_logging(verbose)
-    data = vcf.inspect(icf_path)
+    data = vcf.inspect(path)
     click.echo(tabulate.tabulate(data, headers="keys"))
 
 
 @click.command
-@click.argument("icf_path", type=click.Path())
+@icf_path
 def mkschema(icf_path):
     """
     Generate a schema for zarr encoding
@@ -158,7 +164,7 @@ def mkschema(icf_path):
 
 
 @click.command
-@click.argument("icf_path", type=click.Path())
+@icf_path
 @click.argument("zarr_path", type=click.Path())
 @verbose
 @click.option("-s", "--schema", default=None, type=click.Path(exists=True))
@@ -212,7 +218,7 @@ def encode(
 
 
 @click.command(name="convert")
-@click.argument("vcfs", nargs=-1, required=True)
+@vcfs
 @click.argument("zarr_path", type=click.Path())
 @variants_chunk_size
 @samples_chunk_size
@@ -235,17 +241,6 @@ def convert_vcf(
     )
 
 
-@click.command
-@click.argument("vcfs", nargs=-1, required=True)
-@click.argument("zarr_path", type=click.Path())
-def validate(vcfs, zarr_path):
-    """
-    Development only, do not use. Will be removed before release.
-    """
-    # FIXME! Will silently not look at remaining VCFs
-    vcf.validate(vcfs[0], zarr_path, show_progress=True)
-
-
 @version
 @click.group(cls=NaturalOrderGroup)
 def vcf2zarr():
@@ -309,7 +304,6 @@ def vcf2zarr():
 vcf2zarr.add_command(dexplode_init)
 vcf2zarr.add_command(dexplode_partition)
 vcf2zarr.add_command(dexplode_finalise)
-vcf2zarr.add_command(validate)
 
 
 @click.command(name="convert")

tests/test_cli.py

@@ -9,37 +9,39 @@
 
 
 class TestWithMocks:
+    vcf_path = "tests/data/vcf/sample.vcf.gz"
+
     @mock.patch("bio2zarr.vcf.explode")
-    def test_vcf_explode(self, mocked):
+    def test_vcf_explode(self, mocked, tmp_path):
         runner = ct.CliRunner(mix_stderr=False)
         result = runner.invoke(
-            cli.vcf2zarr, ["explode", "source", "dest"], catch_exceptions=False
+            cli.vcf2zarr, f"explode {self.vcf_path} {tmp_path}", catch_exceptions=False
         )
         assert result.exit_code == 0
         assert len(result.stdout) == 0
         assert len(result.stderr) == 0
         mocked.assert_called_once_with(
-            ("source",),
-            "dest",
+            (self.vcf_path,),
+            str(tmp_path),
             column_chunk_size=64,
             worker_processes=1,
             show_progress=True,
         )
 
-    def test_vcf_dexplode_init(self):
+    def test_vcf_dexplode_init(self, tmp_path):
         runner = ct.CliRunner(mix_stderr=False)
         with mock.patch("bio2zarr.vcf.explode_init", return_value=5) as mocked:
             result = runner.invoke(
                 cli.vcf2zarr,
-                ["dexplode-init", "source", "dest", "5"],
+                f"dexplode-init {self.vcf_path} {tmp_path} 5",
                 catch_exceptions=False,
             )
             assert result.exit_code == 0
             assert len(result.stderr) == 0
             assert result.stdout == "5\n"
             mocked.assert_called_once_with(
-                "dest",
-                ("source",),
+                str(tmp_path),
+                (self.vcf_path,),
                 target_num_partitions=5,
                 worker_processes=1,
                 column_chunk_size=64,
@@ -125,38 +127,21 @@ def test_convert_vcf(self, mocked):
         runner = ct.CliRunner(mix_stderr=False)
         result = runner.invoke(
             cli.vcf2zarr,
-            ["convert", "vcf_path", "zarr_path"],
+            f"convert {self.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",),
+            (self.vcf_path,),
             "zarr_path",
             variants_chunk_size=None,
             samples_chunk_size=None,
             worker_processes=1,
             show_progress=True,
         )
 
-    @mock.patch("bio2zarr.vcf.validate")
-    def test_validate(self, mocked):
-        runner = ct.CliRunner(mix_stderr=False)
-        result = runner.invoke(
-            cli.vcf2zarr,
-            ["validate", "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",
-            show_progress=True,
-        )
-
     @mock.patch("bio2zarr.plink.convert")
     def test_convert_plink(self, mocked):
         runner = ct.CliRunner(mix_stderr=False)
@@ -177,14 +162,14 @@ def test_convert_plink(self, mocked):
 
 
 class TestVcfEndToEnd:
-    data_path = "tests/data/vcf/sample.vcf.gz"
+    vcf_path = "tests/data/vcf/sample.vcf.gz"
 
     def test_dexplode(self, tmp_path):
         icf_path = tmp_path / "icf"
         runner = ct.CliRunner(mix_stderr=False)
         result = runner.invoke(
             cli.vcf2zarr,
-            f"dexplode-init {self.data_path} {icf_path} 5",
+            f"dexplode-init {self.vcf_path} {icf_path} 5",
             catch_exceptions=False,
         )
         assert result.exit_code == 0
@@ -213,7 +198,7 @@ def test_explode(self, tmp_path):
         icf_path = tmp_path / "icf"
         runner = ct.CliRunner(mix_stderr=False)
         result = runner.invoke(
-            cli.vcf2zarr, f"explode {self.data_path} {icf_path}", catch_exceptions=False
+            cli.vcf2zarr, f"explode {self.vcf_path} {icf_path}", catch_exceptions=False
         )
         assert result.exit_code == 0
         result = runner.invoke(
@@ -228,7 +213,7 @@ def test_encode(self, tmp_path):
         zarr_path = tmp_path / "zarr"
         runner = ct.CliRunner(mix_stderr=False)
         result = runner.invoke(
-            cli.vcf2zarr, f"explode {self.data_path} {icf_path}", catch_exceptions=False
+            cli.vcf2zarr, f"explode {self.vcf_path} {icf_path}", catch_exceptions=False
         )
         assert result.exit_code == 0
         result = runner.invoke(
@@ -247,7 +232,7 @@ def test_convert(self, tmp_path):
         runner = ct.CliRunner(mix_stderr=False)
         result = runner.invoke(
             cli.vcf2zarr,
-            f"convert {self.data_path} {zarr_path}",
+            f"convert {self.vcf_path} {zarr_path}",
             catch_exceptions=False,
         )
         assert result.exit_code == 0