Make tests pass. Don't assume /tmp is the OS tmpfile path.

Make seqrepo-rest-service endpoint used in tests configurable with SEQREPO_REST_URL

Add comment to _open_for_reading. Add close to alias db handle

src/biocommons/seqrepo/fastadir/fastadir.py

@@ -38,7 +38,6 @@ class LockableFabgzReader(contextlib.AbstractContextManager):
     def __init__(self, path):
         self.lock = threading.Lock()
         self.fabgz_reader = FabgzReader(path)
-        # super().__init__(*args, **kwargs)
 
     def __enter__(self):
         self.lock.acquire()
@@ -151,18 +150,11 @@ def fetch(self, seq_id, start=None, end=None):
             self.commit()
 
         path = os.path.join(self._root_dir, rec["relpath"])
-        _logger.info(
-            "path: %s, seq_id: %s, start: %s, end: %s",
-            path, seq_id, start, end)
-
-        # fabgz = self._open_for_reading(path)
-        # return fabgz.fetch(seq_id, start, end)
 
         with self._open_for_reading(path) as fabgz:
             seq = fabgz.fetch(seq_id, start, end)
         return seq
 
-
     @functools.lru_cache(maxsize=SEQREPO_LRU_CACHE_MAXSIZE)
     def fetch_seqinfo(self, seq_id):
         """fetch sequence info by seq_id"""
@@ -248,17 +240,17 @@ def _upgrade_db(self):
 
     @functools.lru_cache()
     def _open_for_reading(self, path):
+        """
+        Opens a FabgzReader to path, wraps in a LockableFabgzReader for use in context managers.
+        Places it in an LRU cache so file is only opened once per FastaDir object. Caller must
+        lock the LockableFabgzReader or otherwise handle concurrent access if sharing between
+        in-process concurrent execution threads, such as asyncio (e.g. WSGI/ASGI web servers)
+        """
         _logger.debug("Opening for reading: %s", path)
         if not os.path.exists(path):
             _logger.error("_open_for_reading path does not exist: %s", path)
         return LockableFabgzReader(path)
 
-    # def _open_for_reading(self, path):
-    #     _logger.debug("Opening for reading: " + path)
-    #     if not os.path.exists(path):
-    #         _logger.error("_open_for_reading path does not exist: %s", path)
-    #     return FabgzReader(path)
-
     def _dump_aliases(self):
         import prettytable
 

src/biocommons/seqrepo/seqaliasdb/seqaliasdb.py

@@ -61,6 +61,9 @@ def __init__(
     # ############################################################################
     # Special methods
 
+    def __del__(self):
+        self._db.close()
+
     def __contains__(self, seq_id):
         cursor = self._db.cursor()
         cursor.execute(

tests/conftest.py

@@ -14,7 +14,8 @@ def dataproxy():
 
 @pytest.fixture(scope="session")
 def rest_dataproxy():
-    return SeqRepoRESTDataProxy(base_url="http://localhost:5000/seqrepo")
+    url = os.environ.get("SEQREPO_REST_URL", "http://localhost:5000/seqrepo")
+    return SeqRepoRESTDataProxy(base_url=url)
 
 
 @pytest.fixture(scope="session")

tests/test_fabgz.py

@@ -28,7 +28,7 @@ def test_write_reread():
 
     # now read them back
     far = FabgzReader(fabgz_fn)
-    assert far.filename.startswith("/tmp/".encode())
+    assert far.filename.startswith(tmpdir.encode())
     assert set(far.keys()) == set(sequences.keys())
     assert 5 == len(far), "expected 5 sequences"
     assert "l10" in far.keys()