Merge pull request #191 from GavinHuttley/develop

MAINT: mixed patches

src/ensembl_tui/_align.py

@@ -17,7 +17,7 @@
 
 _no_gaps = numpy.array([], dtype=_DEFAULT_GAP_DTYPE)
 
-ALIGN_STORE_SUFFIX = "align_coords-sqlitedb"
+ALIGN_STORE_SUFFIX = "parquet"
 
 ALIGN_ATTR_SCHEMA = (
     "align_id INTEGER PRIMARY KEY DEFAULT nextval('align_id_seq')",

src/ensembl_tui/_config.py

@@ -200,11 +200,16 @@ def path_to_alignment(self, pattern: str, suffix: str) -> pathlib.Path | None:
         pattern
             glob pattern for the Ensembl alignment name
         """
-        align_dirs = [
-            d
-            for d in self.aligns_path.glob("*")
-            if fnmatch.fnmatch(d.stem, pattern) and d.name.endswith(suffix)
-        ]
+        if eti_util.contains_glob_pattern(pattern):
+            align_dirs = [
+                d
+                for d in self.aligns_path.glob("*")
+                if fnmatch.fnmatch(d.name, pattern)
+            ]
+        elif pattern:
+            align_dirs = [d for d in self.aligns_path.glob("*") if pattern in d.name]
+        else:
+            align_dirs = None
         if not align_dirs:
             return None
 
@@ -214,7 +219,11 @@ def path_to_alignment(self, pattern: str, suffix: str) -> pathlib.Path | None:
                 msg,
             )
 
-        return align_dirs[0]
+        align_dir = align_dirs[0]
+        if not list(align_dir.glob(f"*{suffix}")):
+            msg = f"{align_dir} does not contain file with suffix {suffix}"
+            raise FileNotFoundError(msg)
+        return align_dir
 
 
 def write_installed_cfg(config: Config) -> eti_util.PathType:

src/ensembl_tui/_util.py

@@ -571,3 +571,10 @@ def tempdir(working_dir: pathlib.Path | str | None = None) -> pathlib.Path:
 
 def make_column_constant(schema: tuple[str, ...]) -> tuple[str, ...]:
     return tuple(c.split()[0] for c in schema)
+
+
+_has_wildcard = re.compile(r"[*?\[\]]")
+
+
+def contains_glob_pattern(s: str) -> bool:
+    return _has_wildcard.search(s) is not None

src/ensembl_tui/cli.py

@@ -387,6 +387,7 @@ def species_summary(installed, species):
 @_outdir
 @_align_name
 @_ref
+@_coord_names
 @_ref_genes_file
 @_mask_features
 @_limit
@@ -397,6 +398,7 @@ def alignments(
     outdir,
     align_name,
     ref,
+    coord_names,
     ref_genes_file,
     mask_features,
     limit,
@@ -430,6 +432,14 @@ def alignments(
             text=f"{align_name!r} does not match any alignments under {str(config.aligns_path)!r}",
             colour="red",
         )
+        available = "\n".join(
+            [
+                fn.stem
+                for fn in config.aligns_path.glob("*")
+                if not fn.name.startswith(".") and fn.is_dir()
+            ],
+        )
+        eti_util.print_colour(text=f"Available alignments:\n{available}", colour="red")
         sys.exit(1)
 
     align_db = eti_align.AlignDb(source=align_path)
@@ -463,6 +473,14 @@ def alignments(
             )
             sys.exit(1)
         stableids = table.columns["stableid"]
+    elif coord_names:
+        genome = genomes[ref_species]
+        stableids = list(
+            genome.get_ids_for_biotype(
+                biotype="protein_coding",
+                seqid=coord_names,
+            ),
+        )
     else:
         stableids = None
 

tests/test_config.py

@@ -43,30 +43,45 @@ def test_installed_config_hash():
 @pytest.fixture
 def installed_aligns(tmp_path):
     align_dir = tmp_path / eti_config._COMPARA_NAME / eti_config._ALIGNS_NAME
-    align_dir.mkdir(parents=True, exist_ok=True)
     # make two alignment paths with similar names
-    (align_dir / f"10_primates.epo.{eti_align.ALIGN_STORE_SUFFIX}").open(mode="w")
-    (align_dir / f"24_primates.epo_extended.{eti_align.ALIGN_STORE_SUFFIX}").open(
-        mode="w",
-    )
+    names = "10_primates.epo", "24_primates.epo_extended"
+    for name in names:
+        dirname = align_dir / name
+        dirname.mkdir(parents=True, exist_ok=True)
+        (dirname / f"align_blocks.{eti_align.ALIGN_STORE_SUFFIX}").open(mode="w")
     return eti_config.InstalledConfig(release="11", install_path=tmp_path)
 
 
-@pytest.mark.parametrize("pattern", ("10*", "1*prim*", "10_p*", "10_primates.epo"))
+@pytest.fixture
+def incomplete_installed(installed_aligns):
+    align_path = installed_aligns.aligns_path
+    for path in align_path.glob(
+        f"*/*.{eti_align.ALIGN_STORE_SUFFIX}",
+    ):
+        path.unlink()
+    return installed_aligns
+
+
+@pytest.mark.parametrize("pattern", ["10*", "1*prim*", "10_p*", "10_primates.epo"])
 def test_get_alignment_path(installed_aligns, pattern):
     got = installed_aligns.path_to_alignment(pattern, eti_align.ALIGN_STORE_SUFFIX)
-    assert got.name == f"10_primates.epo.{eti_align.ALIGN_STORE_SUFFIX}"
+    assert got.name == "10_primates.epo"
+
+
+def test_get_alignment_path_incomplete(incomplete_installed):
+    with pytest.raises(FileNotFoundError):
+        incomplete_installed.path_to_alignment("10*", eti_align.ALIGN_STORE_SUFFIX)
 
 
-@pytest.mark.parametrize("pattern", ("10pri*", "blah-blah", ""))
+@pytest.mark.parametrize("pattern", ["10pri*", "blah-blah", ""])
 def test_get_alignment_path_invalid(installed_aligns, pattern):
     assert (
         installed_aligns.path_to_alignment(pattern, eti_align.ALIGN_STORE_SUFFIX)
         is None
     )
 
 
-@pytest.mark.parametrize("pattern", ("*pri*", "*epo*"))
+@pytest.mark.parametrize("pattern", ["*pri*", "*epo*"])
 def test_get_alignment_path_multiple(installed_aligns, pattern):
     with pytest.raises(ValueError):
         installed_aligns.path_to_alignment(pattern, eti_align.ALIGN_STORE_SUFFIX)