Merge pull request #2 from GavinHuttley/develop

MAINT: merge develop changes

.gitignore

@@ -50,3 +50,6 @@ venv/
 working/
 # vi
 .*.swp
+
+# test data installed via download
+small-113/*

.hgignore

@@ -35,4 +35,5 @@ doc/draw*
 dist/*
 working/*
 lcov*.info
-.ruff_cache/*
+.ruff_cache/*
+small-113/*

src/ensembl_tui/_align.py

@@ -1,5 +1,6 @@
 import pathlib
 import typing
+import uuid
 from collections import defaultdict
 from dataclasses import dataclass
 
@@ -85,6 +86,9 @@ def __init__(
         mode: str = "r",
         in_memory: bool = False,
     ):
+        in_memory = in_memory or "memory" in str(source)
+        source = uuid.uuid4().hex if in_memory else source
+
         self.source = pathlib.Path(source)
         self.mode = "w-" if mode == "w" else mode
         h5_kwargs = (

src/ensembl_tui/_download.py

@@ -171,7 +171,7 @@ class valid_compara_homology:
     """homology tsv files"""
 
     def __init__(self) -> None:
-        self._valid = re.compile("([.]tsv[.]gz|README|MD5SUM)")
+        self._valid = re.compile("([.]tsv|[.]tsv[.]gz|README|MD5SUM)$")
 
     def __call__(self, name: str) -> bool:
         return self._valid.search(name) is not None
@@ -208,7 +208,7 @@ def download_homology(
             elt_ftp.listdir(config.host, remote_path, valid_compara_homology()),
         )
         if verbose:
-            print(remote_paths)
+            print(f"{remote_path=}", f"{remote_paths=}", sep="\n")
 
         if debug:
             # we need the checksum files

src/ensembl_tui/_genome.py

@@ -6,6 +6,7 @@
 import re
 import sqlite3
 import typing
+import uuid
 from abc import ABC, abstractmethod
 from typing import Any
 
@@ -799,25 +800,25 @@ def __init__(
         in_memory: bool = False,
     ):
         # note that species are converted into the Ensembl db prefix
+        in_memory = in_memory or "memory" in str(source)
+        source = uuid.uuid4().hex if in_memory else source
+        self.source = pathlib.Path(source)
 
-        source = pathlib.Path(source)
-        self.source = source
-
-        if mode == "r" and not source.exists():
+        if not in_memory and mode == "r" and not self.source.exists():
             raise OSError(f"{self.source!s} not found")
 
         species = (
             elt_species.Species.get_ensembl_db_prefix(species) if species else None
         )
         self.mode = "w-" if mode == "w" else mode
-        if in_memory:
-            h5_kwargs = dict(
+        h5_kwargs = (
+            dict(
                 driver="core",
                 backing_store=False,
             )
-        else:
-            h5_kwargs = {}
-
+            if in_memory
+            else {}
+        )
         try:
             self._file: h5py.File = h5py.File(source, mode=self.mode, **h5_kwargs)
         except OSError:

src/ensembl_tui/_install.py

@@ -176,7 +176,7 @@ def local_install_homology(
     dirnames = []
     for sp in config.db_names:
         path = config.staging_homologies / sp
-        dirnames.extend(list(path.glob("*.tsv.gz")))
+        dirnames.extend(list(path.glob("*.tsv*")))
 
     if max_workers:
         max_workers = min(len(dirnames) + 1, max_workers)