purge_old_runs() now takes a prefix parameter, to be more certain that the data being deleted, really are intended to be deleted.

src/purge/purge_old_runs.py

@@ -7,9 +7,9 @@ import time
 
 logger = logging.getLogger(__name__)
 
-def purge_old_runs(run_dir: str, max_age_hours: float = 1.0) -> None:
+def purge_old_runs(run_dir: str, max_age_hours: float = 1.0, prefix: str = "myrun_") -> None:
     """
-    Deletes runs in the specified run_dir older than max_age_hours.
+    Deletes runs in the specified run_dir older than max_age_hours and matching the specified prefix.
     """
     if not os.path.isabs(run_dir):
         raise ValueError(f"run_dir must be an absolute path: {run_dir}")
@@ -18,6 +18,9 @@ def purge_old_runs(run_dir: str, max_age_hours: float = 1.0) -> None:
     cutoff = now - datetime.timedelta(hours=max_age_hours)
 
     for run_id in os.listdir(run_dir):
+        if not run_id.startswith(prefix):
+            continue  # Skip files and directories that don't match the prefix
+
         run_path = os.path.join(run_dir, run_id)
         if not os.path.isdir(run_path):
             continue  # Skip files
@@ -35,7 +38,7 @@ def purge_old_runs(run_dir: str, max_age_hours: float = 1.0) -> None:
         except Exception as e:
             logger.error(f"Error processing {run_id} in {run_dir}: {e}")
 
-def start_purge_scheduler(run_dir: str, purge_interval_seconds: float=3600) -> None:
+def start_purge_scheduler(run_dir: str, purge_interval_seconds: float=3600, prefix: str = "myrun_") -> None:
     """
     Start the purge scheduler in a background thread.
     """
@@ -51,7 +54,7 @@ def start_purge_scheduler(run_dir: str, purge_interval_seconds: float=3600) -> N
         while True:
             logger.info("Running purge...")
             print("Running purge...")
-            purge_old_runs(run_dir)
+            purge_old_runs(run_dir, prefix=prefix)
             time.sleep(purge_interval_seconds)
 
     purge_thread = threading.Thread(target=purge_scheduler, daemon=True)

src/purge/tests/test_purge_old_runs.py

@@ -6,9 +6,7 @@ from src.purge.purge_old_runs import purge_old_runs
 
 class TestPurgeOldRuns(unittest.TestCase):
     def setUp(self):
-        """
-        Set up test environment before each test.
-        """
+        """Set up test environment before each test."""
         # Create a temporary directory for the runs
         self.test_run_dir = os.path.abspath(os.path.join(os.path.dirname(__file__), "test_run"))
         if os.path.exists(self.test_run_dir):
@@ -16,26 +14,23 @@ class TestPurgeOldRuns(unittest.TestCase):
         os.makedirs(self.test_run_dir, exist_ok=True)
 
         # Create some dummy run directories with different modification times
-        self.create_dummy_run("run1", hours_old=0.5)
-        self.create_dummy_run("run2", hours_old=1.5)  # Should be purged
-        self.create_dummy_run("run3", hours_old=2) # Should be purged
-        self.create_dummy_run("run4", hours_old=0.25)
-        self.create_dummy_run("run5", hours_old=1)  # Boundary condition, might be purged
-        self.create_dummy_run("run6", hours_old=0)  # Today
+        self.create_dummy_run("myrun_run1", hours_old=0.5)
+        self.create_dummy_run("myrun_run2", hours_old=1.5)  # Should be purged
+        self.create_dummy_run("myrun_run3", hours_old=2) # Should be purged
+        self.create_dummy_run("myrun_run4", hours_old=0.25)
+        self.create_dummy_run("myrun_run5", hours_old=1)  # Boundary condition, might be purged
+        self.create_dummy_run("myrun_run6", hours_old=0)  # Today
+        self.create_dummy_run("other_run7", hours_old=1.5) #Should NOT be purged
         self.create_dummy_file("not_a_run.txt") # a file that should be left alone
 
     def tearDown(self):
-        """
-        Clean up test environment after each test.
-        """
+        """Clean up test environment after each test."""
         # Remove the temporary run directory and its contents
         if os.path.exists(self.test_run_dir):
             shutil.rmtree(self.test_run_dir)
 
     def create_dummy_run(self, run_id, hours_old):
-        """
-        Creates a dummy run directory with a specific modification time.
-        """
+        """Creates a dummy run directory with a specific modification time."""
         run_path = os.path.join(self.test_run_dir, run_id)
         os.makedirs(run_path, exist_ok=True)
 
@@ -52,11 +47,11 @@ class TestPurgeOldRuns(unittest.TestCase):
     def test_purge_old_runs(self):
         """Tests the purge_old_runs function."""
         max_age_hours = 0.95
-        purge_old_runs(self.test_run_dir, max_age_hours=max_age_hours)  # Pass the directory
+        purge_old_runs(self.test_run_dir, max_age_hours=max_age_hours, prefix="myrun_")  # Pass the directory
 
         # Check which runs should have been purged
-        runs_to_keep = ["run1", "run4", "run6", "not_a_run.txt"]
-        runs_to_purge = ["run2", "run3", "run5"]
+        runs_to_keep = ["myrun_run1", "myrun_run4", "myrun_run6", "other_run7","not_a_run.txt"]
+        runs_to_purge = ["myrun_run2", "myrun_run3", "myrun_run5"]
 
         for run_id in runs_to_keep:
             run_path = os.path.join(self.test_run_dir, run_id)
@@ -76,10 +71,10 @@ class TestPurgeOldRuns(unittest.TestCase):
           os.utime(item_path, (mtime, mtime))
 
         max_age_hours = 1.0
-        purge_old_runs(self.test_run_dir, max_age_hours=max_age_hours)  # Pass the directory
+        purge_old_runs(self.test_run_dir, max_age_hours=max_age_hours, prefix="myrun_")  # Pass the directory
 
-        # All runs should still exist.
-        expected_runs = ["run1", "run2", "run3", "run4", "run5", "run6", "not_a_run.txt"]
+        # All runs should still exist, including the one with the wrong prefix.
+        expected_runs = ["myrun_run1", "myrun_run2", "myrun_run3", "myrun_run4", "myrun_run5", "myrun_run6", "other_run7", "not_a_run.txt"]
         for run_id in expected_runs:
             run_path = os.path.join(self.test_run_dir, run_id)
             self.assertTrue(os.path.exists(run_path), f"Run {run_id} should not have been purged.")