Merge pull request #50 from gbouras13/dev

v1.6.2

Author: gbouras13

HISTORY.md

@@ -1,5 +1,11 @@
 # History
 
+1.6.2 (2024-03-08)
+------------------
+
+* Bug fix with `--force` for `plassembler download`, which will only remove the output directory if `--force` is specified (#49).
+
+
 1.6.1 (2024-03-02)
 ------------------
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "plassembler"
-version = "1.6.1" # change VERSION too
+version = "1.6.2" # change VERSION too
 description = "Quickly and accurately assemble plasmids in hybrid sequenced bacterial isolates"
 authors = ["George Bouras <george.bouras@adelaide.edu.au>"]
 license = "MIT"

src/plassembler/__init__.py

@@ -434,7 +434,7 @@ def run(
     # check the mash database is installed
 
     logger.info("Checking database installation.")
-    check_db_installation(Path(database), install_flag=False)
+    check_db_installation(Path(database), force=False, install_flag=False)
     # will only continue if successful
     logger.info("Database successfully checked.")
 
@@ -996,7 +996,7 @@ def assembled(
     # check the mash database is installed
 
     logger.info("Checking database installation.")
-    check_db_installation(Path(database), install_flag=False)
+    check_db_installation(Path(database), force=False, install_flag=False)
     # will only continue if successful
     logger.info("Database successfully checked.")
 
@@ -1121,7 +1121,7 @@ def download(ctx, database, force, **kwargs):
     logger.add(lambda _: sys.exit(1), level="ERROR")
     database = Path(database)
     logger.info(f"Checking database installation at {database}")
-    check_db_installation(database, install_flag=True)  # t
+    check_db_installation(database, force, install_flag=True)  # t
 
 
 """
@@ -1319,7 +1319,7 @@ def long(
 
     # check the mash database is installed
     logger.info("Checking database installation.")
-    check_db_installation(Path(database), install_flag=False)
+    check_db_installation(Path(database), force=False, install_flag=False)
     # will only continue if successful
     logger.info("Database successfully checked.")
 

src/plassembler/utils/VERSION

@@ -1 +1 @@
-1.6.1
+1.6.2

src/plassembler/utils/db.py

@@ -16,9 +16,10 @@
 from plassembler.utils.cleanup import remove_directory
 
 
-def check_db_installation(db_dir: Path, install_flag: bool):
+def check_db_installation(db_dir: Path, force: bool, install_flag: bool):
     """checks database is installed correctly
     :param db_dir: database directory
+    :param force: boolean if force is true
     :param install_flag: whether to check or install database
     """
     # Mash files
@@ -28,6 +29,25 @@ def check_db_installation(db_dir: Path, install_flag: bool):
     f1: Path = db_dir / f"{mash_db_names[0]}"
     f2: Path = db_dir / f"{mash_db_names[1]}"
 
+    if install_flag is True:
+        if force is True:
+            if os.path.isdir(db_dir) is True:
+                logger.info(f"Removing the directory {db_dir} as --force was specified")
+                shutil.rmtree(db_dir)
+            else:
+                logger.info(
+                    f"--force was specified even though the directory {db_dir} does not already exist. Continuing"
+                )
+        else:
+            if os.path.isdir(db_dir) is True:
+                logger.error(
+                    f"Directory {db_dir} already exists and force was not specified. Please specify -f or --force to overwrite {db_dir}"
+                )
+
+    # instantiate outdir
+    if os.path.isdir(db_dir) is False:
+        os.mkdir(db_dir)
+
     if f1.exists() and f2.exists():
         logger.info(f"PLSDB Database mash sketch at {f1} exists.")
         logger.info(f"PLSDB Database tsv metadata file at {f2} exists.")
@@ -49,20 +69,12 @@ def check_db_installation(db_dir: Path, install_flag: bool):
 
 
 def get_database_zenodo(db_dir: Path):
-    logger.info("Downloading Plassembler Database.")
+    logger.info(f"Downloading Plassembler Database to the directory {db_dir}")
     tarball = "201123_plassembler_v1.5.0_databases.tar.gz"
     tar_path = Path(f"{db_dir}/{tarball}")
     db_url = "https://zenodo.org/record/10158040/files/201123_plassembler_v1.5.0_databases.tar.gz"
     requiredmd5 = "3a24bacc05bb857dc044fc6662b58db7"
 
-    # remvoe the directory
-    if os.path.exists(db_dir):
-        shutil.rmtree(db_dir)
-
-    # make db dir
-    if not os.path.exists(db_dir):
-        os.mkdir(db_dir)
-
     try:
         with tar_path.open("wb") as fh_out, requests.get(db_url, stream=True) as resp:
             total_length = resp.headers.get("content-length")

tests/test_db.py

@@ -40,19 +40,12 @@ class test_install(unittest.TestCase):
 
     # for plassembler run
     def test_check_db_installation_good(self):
-        check_db_installation(db_path, False)
+        check_db_installation(db_path, force=False, install_flag=False)
 
     # for plassembler download
     def test_check_db_installation_good_d(self):
-        check_db_installation(db_path, True)
+        check_db_installation(db_path, force=True, install_flag=True)
 
     def test_check_db_installation_bad(self):
         with self.assertRaises(SystemExit):
-            check_db_installation(val_data, False)
-
-    def test_get_database_zenodo(self):
-        expected_return = True
-        get_database_zenodo(tmp_db_path)
-        # remove it after downloading
-        shutil.rmtree(tmp_db_path)
-        self.assertEqual(expected_return, True)
+            check_db_installation(val_data, force=False, install_flag=False)