fix(scrubbing): Catch exceptions when download required spacy model (#733)

Author: KIRA009

openadapt/app/cards.py

@@ -5,7 +5,6 @@
 
 from datetime import datetime
 import multiprocessing
-import sys
 import time
 
 from nicegui import ui
@@ -42,17 +41,11 @@ def reset(self) -> None:
 
     def wait(self) -> None:
         """Wait for the recording process to finish."""
-        if sys.platform == "win32":
-            # a bug on windows for when `record` runs for a long time, where even when
-            # the `record` function returns a value, `record_proc` refuses to join.
-            # this is a workaround to ensure the process is terminated
-            while True:
-                if self.terminate_recording.is_set():
-                    self.record_proc.terminate()
-                    return
-                time.sleep(0.1)
-        else:
-            self.record_proc.join()
+        while True:
+            if self.terminate_recording.is_set():
+                self.record_proc.terminate()
+                return
+            time.sleep(0.1)
 
     def is_running(self) -> bool:
         """Check if the recording process is running."""

openadapt/app/dashboard/app/scrubbing/ScrubbingUpdates.tsx

@@ -25,7 +25,7 @@ export const ScrubbingUpdates = ({ data, resetScrubbingStatus }: Props) => {
         <Text component='h2'>Scrubbing updates for recording {data.recording.task_description}</Text>
         <Text>Provider: {data.provider}</Text>
         {data.copying_recording ? (
-          <Text>Copying recording first...</Text>
+          <Text>Copying recording (this may take a while if Spacy dependencies need to be downloaded on the first run)...</Text>
         ) : data.error ? (
           <Text c="red">{data.error}</Text>
         ) : (

openadapt/privacy/providers/presidio.py

@@ -22,13 +22,6 @@
 from openadapt.privacy.base import Modality, ScrubbingProvider, TextScrubbingMixin
 from openadapt.privacy.providers import ScrubProvider
 
-if not spacy.util.is_package(config.SPACY_MODEL_NAME):  # pylint: disable=no-member
-    logger.info(f"Downloading {config.SPACY_MODEL_NAME} model...")
-    if not is_running_from_executable():
-        spacy.cli.download(config.SPACY_MODEL_NAME)
-    else:
-        download_spacy_model(config.SPACY_MODEL_NAME)
-
 
 class PresidioScrubbingProvider(
     ScrubProvider, ScrubbingProvider, TextScrubbingMixin
@@ -38,6 +31,13 @@ class PresidioScrubbingProvider(
     name: str = ScrubProvider.PRESIDIO  # pylint: disable=E1101
     capabilities: List[Modality] = [Modality.TEXT, Modality.PIL_IMAGE]
 
+    if not spacy.util.is_package(config.SPACY_MODEL_NAME):  # pylint: disable=no-member
+        logger.info(f"Downloading {config.SPACY_MODEL_NAME} model...")
+        if not is_running_from_executable():
+            spacy.cli.download(config.SPACY_MODEL_NAME)
+        else:
+            download_spacy_model(config.SPACY_MODEL_NAME)
+
     def scrub_text(self, text: str, is_separated: bool = False) -> str:
         """Scrub the text of all PII/PHI using Presidio ANALYZER.TRF and Anonymizer.
 

openadapt/scrub.py

@@ -163,7 +163,15 @@ def inner() -> None:
 
         new_recording = crud.get_recording_by_id(write_session, new_recording_id)
 
-        scrubber = ScrubProvider.get_scrubber(provider_id)
+        try:
+            scrubber = ScrubProvider.get_scrubber(provider_id)
+        except Exception as e:
+            logger.error(e)
+            cleanup(
+                "Failed to download the required spacy model. Please try again after"
+                " some time."
+            )
+            return
 
         crud.scrub_item(new_recording_id, Recording, scrubber)