Merge pull request #26 from jeromekelleher/yet-more-cli

Yet more cli

Author: jeromekelleher

biofuse/cli.py

@@ -63,7 +63,7 @@ def biofuse_main():
     """biofuse: read-only FUSE filesystem views over VCF Zarr data."""
 
 
-@biofuse_main.command(name="mount-plink")
+@biofuse_main.command(name="mount-plink", cls=vcztools.GroupedCommand)
 @click.argument("vcz_url", type=str)
 @click.argument(
     "mount_dir",
@@ -88,13 +88,11 @@ def mount_plink(vcz_url, mount_dir, basename, access_log_path, **kwargs):
 
     The mount runs in the foreground until interrupted with Ctrl-C.
     """
-    opts = vcztools.ViewPlinkOptions.pop_from_click_kwargs(kwargs)
-    _run_mount(
-        formats.PLINK_SPEC, vcz_url, mount_dir, basename, access_log_path, opts, kwargs
-    )
+    opts = vcztools.ViewPlinkOptions.from_click_kwargs(kwargs)
+    _run_mount(formats.PLINK_SPEC, vcz_url, mount_dir, basename, access_log_path, opts)
 
 
-@biofuse_main.command(name="mount-bgen")
+@biofuse_main.command(name="mount-bgen", cls=vcztools.GroupedCommand)
 @click.argument("vcz_url", type=str)
 @click.argument(
     "mount_dir",
@@ -119,15 +117,12 @@ def mount_bgen(vcz_url, mount_dir, basename, access_log_path, **kwargs):
     from ``vcztools view-bgen``; see ``vcztools view-bgen --help`` for the
     full reference. The encoder always uses zlib level 0 (stored,
     fixed-size blocks) so byte-range random access into the mounted
-    ``.bgen`` is O(1); ``--compression-level`` is accepted for parity
-    with ``view-bgen`` but has no effect.
+    ``.bgen`` is O(1).
 
     The mount runs in the foreground until interrupted with Ctrl-C.
     """
-    opts = vcztools.ViewBgenOptions.pop_from_click_kwargs(kwargs)
-    _run_mount(
-        formats.BGEN_SPEC, vcz_url, mount_dir, basename, access_log_path, opts, kwargs
-    )
+    opts = vcztools.ViewBgenOptions.from_click_kwargs(kwargs)
+    _run_mount(formats.BGEN_SPEC, vcz_url, mount_dir, basename, access_log_path, opts)
 
 
 def _run_mount(
@@ -137,9 +132,7 @@ def _run_mount(
     basename: str | None,
     access_log_path: str | None,
     opts,
-    extra_kwargs: dict,
 ) -> None:
-    assert extra_kwargs == {}, extra_kwargs
     opts.log.apply()
 
     mount_dir_path = pathlib.Path(mount_dir)

biofuse/formats.py

@@ -27,20 +27,18 @@
 sample identifiers from its header block.
 
 For BGEN the ``.bgen.bgi`` sidecar is a SQLite database that
-:func:`vcztools.bgen.write_bgen_index` writes to a filesystem path.
-We materialise it to a tempfile at session-init time, read the bytes
-back, and hold them in the server process's memory alongside the
-``.sample`` text.
+:func:`vcztools.write_bgi` writes to a filesystem path. We materialise
+it to a tempfile at session-init time, read the bytes back, and hold
+them in the server process's memory alongside the ``.sample`` text.
 """
 
 import dataclasses
+import io
 import pathlib
 import tempfile
 from collections.abc import Callable
 
 import vcztools
-from vcztools import bgen as vcztools_bgen
-from vcztools import plink as vcztools_plink
 
 
 @dataclasses.dataclass(frozen=True)
@@ -86,9 +84,13 @@ def _plink_static_suffixes(opts) -> tuple[str, ...]:
 def _build_plink_static(reader, opts) -> dict[str, bytes]:
     out: dict[str, bytes] = {}
     if not opts.no_bim:
-        out[".bim"] = vcztools_plink.generate_bim(reader).encode("utf-8")
+        buf = io.StringIO()
+        vcztools.write_bim(reader, buf)
+        out[".bim"] = buf.getvalue().encode("utf-8")
     if not opts.no_fam:
-        out[".fam"] = vcztools_plink.generate_fam(reader).encode("utf-8")
+        buf = io.StringIO()
+        vcztools.write_fam(reader, buf)
+        out[".fam"] = buf.getvalue().encode("utf-8")
     return out
 
 
@@ -104,18 +106,20 @@ def _bgen_static_suffixes(opts) -> tuple[str, ...]:
 def _build_bgen_static(reader, opts) -> dict[str, bytes]:
     out: dict[str, bytes] = {}
     if not opts.no_sample_file:
-        out[".sample"] = vcztools_bgen.generate_sample(reader).encode("utf-8")
+        buf = io.StringIO()
+        vcztools.write_sample(reader, buf)
+        out[".sample"] = buf.getvalue().encode("utf-8")
     if not opts.no_bgi:
-        # ``write_bgen_index`` takes a filesystem path. Materialise the
-        # .bgi into a TemporaryDirectory, read the bytes back, then let
-        # the context manager clean up. The encoder used to harvest
-        # ``variant_offsets`` is I/O-free in ``__init__`` and is
-        # discarded once the offsets are read.
+        # ``write_bgi`` requires a filesystem path (sqlite3.connect needs
+        # a real path). Materialise the .bgi into a TemporaryDirectory,
+        # read the bytes back, then let the context manager clean up.
+        # The encoder used to harvest ``variant_offsets`` is I/O-free in
+        # ``__init__`` and is discarded once the offsets are read.
         with vcztools.BgenEncoder(reader) as encoder:
             variant_offsets = encoder.variant_offsets
         with tempfile.TemporaryDirectory(prefix="biofuse-bgen-") as tmp_dir:
             bgi_path = pathlib.Path(tmp_dir) / "index.bgen.bgi"
-            vcztools_bgen.write_bgen_index(reader, str(bgi_path), variant_offsets)
+            vcztools.write_bgi(reader, str(bgi_path), variant_offsets)
             out[".bgen.bgi"] = bgi_path.read_bytes()
     return out
 

tests/test_cli.py

@@ -35,6 +35,13 @@ def test_mount_plink_help(self):
             "--log-level",
         ]:
             assert flag in result.output, f"missing {flag} in mount-plink help"
+        for section in [
+            "Selection options:",
+            "Zarr store options:",
+            "Reader options:",
+            "Logging options:",
+        ]:
+            assert section in result.output, f"missing {section} in mount-plink help"
 
     def test_mount_bgen_help(self):
         result = CliRunner().invoke(cli.biofuse_main, ["mount-bgen", "--help"])
@@ -47,10 +54,16 @@ def test_mount_bgen_help(self):
             "--no-sample-file",
             "--no-bgi",
             "--no-header-samples",
-            "--compression-level",
             "--log-level",
         ]:
             assert flag in result.output, f"missing {flag} in mount-bgen help"
+        for section in [
+            "Selection options:",
+            "Zarr store options:",
+            "Reader options:",
+            "Logging options:",
+        ]:
+            assert section in result.output, f"missing {section} in mount-bgen help"
 
     def test_version(self):
         result = CliRunner().invoke(cli.biofuse_main, ["--version"])

tests/test_formats.py

@@ -7,17 +7,22 @@
 the same selection.
 """
 
+import io
 import sqlite3
 import tempfile
 
 import pytest
 import vcztools
-from vcztools import bgen as vcztools_bgen
-from vcztools import plink as vcztools_plink
 
 from biofuse import formats
 
 
+def _write_to_string(writer, reader):
+    buf = io.StringIO()
+    writer(reader, buf)
+    return buf.getvalue().encode("utf-8")
+
+
 def _open_reader(path):
     return vcztools.ViewPlinkOptions().make_reader(str(path))
 
@@ -54,11 +59,11 @@ def test_identifiers(self, fx_plink_opts):
         assert formats.PLINK_SPEC.streaming_kind == "bed"
         assert formats.PLINK_SPEC.static_suffixes(fx_plink_opts) == (".bim", ".fam")
 
-    def test_static_files_match_generators(self, fx_reader, fx_plink_opts):
+    def test_static_files_match_writers(self, fx_reader, fx_plink_opts):
         static = formats.PLINK_SPEC.build_static_files(fx_reader, fx_plink_opts)
         assert set(static) == {".bim", ".fam"}
-        assert static[".bim"] == vcztools_plink.generate_bim(fx_reader).encode("utf-8")
-        assert static[".fam"] == vcztools_plink.generate_fam(fx_reader).encode("utf-8")
+        assert static[".bim"] == _write_to_string(vcztools.write_bim, fx_reader)
+        assert static[".fam"] == _write_to_string(vcztools.write_fam, fx_reader)
 
     def test_encoder_total_size(self, fx_reader, fx_small_vcz, fx_plink_opts):
         with formats.PLINK_SPEC.encoder_factory(fx_reader, fx_plink_opts) as encoder:
@@ -81,11 +86,9 @@ def test_static_files_keys_match_suffixes(self, fx_reader, fx_bgen_opts):
         static = formats.BGEN_SPEC.build_static_files(fx_reader, fx_bgen_opts)
         assert set(static) == {".sample", ".bgen.bgi"}
 
-    def test_sample_matches_generator(self, fx_reader, fx_bgen_opts):
+    def test_sample_matches_writer(self, fx_reader, fx_bgen_opts):
         static = formats.BGEN_SPEC.build_static_files(fx_reader, fx_bgen_opts)
-        assert static[".sample"] == vcztools_bgen.generate_sample(fx_reader).encode(
-            "utf-8"
-        )
+        assert static[".sample"] == _write_to_string(vcztools.write_sample, fx_reader)
 
     def test_bgi_opens_as_sqlite(self, fx_reader, tmp_path, fx_small_vcz, fx_bgen_opts):
         static = formats.BGEN_SPEC.build_static_files(fx_reader, fx_bgen_opts)