Merge pull request #297 from GEECS-BELLA/feat/post-analysis-cleanup

Add post-analysis cleanup hook to free per-scan memory

Author: skbarber

ScanAnalysis/scan_analysis/analyzers/common/scatter_plotter_analysis.py

@@ -489,3 +489,7 @@ def process_to_bins(self, key_name: str) -> dict[str, np.ndarray]:
                 logging.info(f"Shots: {shots_in_bin[0]} - {shots_in_bin[-1]}")
 
         return {"bin": bins, "average": average, "sigma": sigma, "median": median}
+
+    def cleanup(self) -> None:
+        """No large data held between scans; nothing to release."""
+        pass

ScanAnalysis/scan_analysis/analyzers/common/single_device_scan_analyzer.py

@@ -815,6 +815,30 @@ def bin_data_from_results(self) -> dict[int, BinDataEntry]:
 
         return binned_data
 
+    def cleanup(self) -> None:
+        """
+        Free memory held by loaded and analyzed data after analysis is complete.
+
+        Clears per-scan attributes that may hold large numpy arrays or result objects:
+
+        - ``raw_data`` — loaded shot images
+        - ``results`` — per-shot/bin ImageAnalyzerResult objects
+        - ``_data_file_map`` — shot-to-path mapping
+        - ``stateful_results`` — batch-analysis state dict
+        - ``_pending_aux_updates`` — queued s-file row updates
+
+        Also delegates to ``renderer.cleanup()``.
+        """
+        self.raw_data = {}
+        self.results = {}
+        self._data_file_map = {}
+        self.stateful_results = {}
+        self._pending_aux_updates = []
+
+        self.renderer.cleanup()
+
+        logger.debug(f"[{self.__class__.__name__}] cleanup() complete.")
+
     @staticmethod
     def average_data(data_list: list[np.ndarray]) -> Optional[np.ndarray]:
         """Return the element-wise mean of a list of data arrays."""

ScanAnalysis/scan_analysis/analyzers/renderers/base_renderer.py

@@ -1,10 +1,13 @@
 """Abstract base class for scan analysis renderers."""
 
+import logging
 from abc import ABC, abstractmethod
 from pathlib import Path
 from typing import List
 from .config import RenderContext, BaseRendererConfig
 
+logger = logging.getLogger(__name__)
+
 
 class BaseRenderer(ABC):
     """
@@ -113,3 +116,14 @@ def render_animation(
             Path to the created animation file
         """
         pass
+
+    def cleanup(self) -> None:
+        """
+        Free memory held by the renderer after analysis is complete.
+
+        Resets ``display_contents`` so paths from a previous scan are not
+        accidentally returned on the next run.  Subclasses may override to
+        release additional cached state (e.g. large arrays held across calls).
+        """
+        self.display_contents = []
+        logger.debug(f"[{self.__class__.__name__}] cleanup() complete.")

ScanAnalysis/scan_analysis/base.py

@@ -138,6 +138,17 @@ def run_analysis(self, scan_tag: ScanTag) -> Optional[list[Union[Path, str]]]:
             return None
         return self._run_analysis_core()
 
+    def cleanup(self) -> None:
+        """Release per-scan memory after analysis completes.
+
+        Must be implemented by all subclasses. Called by the task runner
+        after each analyzer finishes so that loaded data and results do not
+        accumulate across scans. Failing to implement this will raise
+        NotImplementedError and halt the runner — intentionally, to prevent
+        unbounded memory growth.
+        """
+        raise NotImplementedError(f"{self.__class__.__name__} must implement cleanup()")
+
     def _run_analysis_core(self) -> Optional[list[Union[Path, str]]]:
         """Core analysis routine to be implemented by subclasses."""
         """

ScanAnalysis/scan_analysis/task_queue.py

@@ -433,6 +433,7 @@ def run_worklist(
         finally:
             stop_event.set()
             hb_thread.join(timeout=HEARTBEAT_INTERVAL_SECONDS)
+            analyzer.cleanup()
 
 
 def load_analyzers_from_config(

docs/scan_analysis/examples/live_watch.ipynb

@@ -41,12 +41,13 @@
     "scan_analysis_config.set_base_dir(ScanPaths.paths_config.scan_analysis_configs_path)\n",
     "\n",
     "\n",
-    "date_tag = ScanTag(year=2026, month=2, day=12, number=0, experiment=\"Thomson\")\n",
+    "date_tag = ScanTag(year=2026, month=2, day=12, number=0, experiment=\"Undulator\")\n",
     "runner = LiveTaskRunner(\n",
-    "    analyzer_group=\"HTT\",\n",
+    "    analyzer_group=\"Undulator\",\n",
     "    date_tag=date_tag,\n",
     "    config_dir=scan_analysis_config.base_dir,  # optional; else env/global\n",
     "    image_config_dir=image_analysis_config.base_dir,  # optional\n",
+    "    gdoc_enabled=True,\n",
     ")"
    ]
   },
@@ -83,7 +84,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3 (ipykernel)",
+   "display_name": "Python (Poetry)",
    "language": "python",
    "name": "python3"
   },
@@ -97,7 +98,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.18"
+   "version": "3.10.11"
   }
  },
  "nbformat": 4,