trying to fix the scrubbing configurations

openadapt/config.py

@@ -5,9 +5,8 @@
 from dotenv import load_dotenv
 from loguru import logger
 from presidio_analyzer import AnalyzerEngine
-from presidio_analyzer.nlp_engine import NlpEngineProvider
 from presidio_anonymizer import AnonymizerEngine
-from presidio_image_redactor import ImageRedactorEngine, ImageAnalyzerEngine
+from presidio_image_redactor import ImageRedactorEngine
 
 
 _DEFAULTS = {
@@ -51,18 +50,25 @@ def getenv_fallback(var_name):
 
 # SCRUBBING CONFIGURATIONS
 
-SCRUB_CONFIG = {
-    "nlp_engine_name": "spacy",
-    "models": [{"lang_code": "en", "model_name": "en_core_web_trf"}],
-}
-SCRUB_PROVIDER = NlpEngineProvider(nlp_configuration=SCRUB_CONFIG)
-NLP_ENGINE = SCRUB_PROVIDER.create_engine()
-ANALYZER = AnalyzerEngine(
-    nlp_engine=NLP_ENGINE,
-    supported_languages=["en"]
-)
+# SCRUB_CONFIG_TRF = {
+#     "nlp_engine_name": "spacy",
+#     "models": [{"lang_code": "en", "model_name": "en_core_web_trf"}],
+# }
+# SCRUB_PROVIDER_TRF = NlpEngineProvider(nlp_configuration=SCRUB_CONFIG_TRF)
+# NLP_ENGINE_TRF = SCRUB_PROVIDER_TRF.create_engine()
+# SCRUB_CONFIG_LG = {
+#     "nlp_engine_name": "spacy",
+#     "models": [{"lang_code": "en", "model_name": "en_core_web_lg"}],
+# }
+# SCRUB_PROVIDER_LG = NlpEngineProvider(nlp_configuration=SCRUB_CONFIG_LG)
+# NLP_ENGINE_LG = SCRUB_PROVIDER_LG.create_engine()
+# ANALYZER = AnalyzerEngine(
+#     nlp_engine=NLP_ENGINE_TRF,
+#     supported_languages=["en"]
+# )
+ANALYZER = AnalyzerEngine()
 ANONYMIZER = AnonymizerEngine()
-IMAGE_REDACTOR = ImageRedactorEngine(ImageAnalyzerEngine(ANALYZER))
+IMAGE_REDACTOR = ImageRedactorEngine()
 SCRUB_IGNORE_ENTITIES = [
     # 'US_PASSPORT',
     # 'US_DRIVER_LICENSE',

openadapt/record.py

@@ -382,13 +382,6 @@ def read_window_events(
             window_data["title"] != prev_window_data.get("title") or
             window_data["window_id"] != prev_window_data.get("window_id")
         ):
-            # TODO: fix exception sometimes triggered by the next line on win32:
-            #   File "\Python39\lib\threading.py" line 917, in run
-            #   File "...\openadapt\record.py", line 277, in read window events
-            #   File "...\env\lib\site-packages\loguru\logger.py" line 1977, in info
-            #   File "...\env\lib\site-packages\loguru\_logger.py", line 1964, in _log
-            #       for handler in core.handlers.values):
-            #   RuntimeError: dictionary changed size during iteration
             _window_data = dict(window_data)
             _window_data.pop("state")
             logger.info(f"{_window_data=}")
@@ -399,7 +392,7 @@ def read_window_events(
                 "window",
                 window_data,
             ))
-        prev_window_data = window_data
+        prev_window_data = dict(window_data)
 
 
 def performance_stats_writer (

openadapt/utils.py

@@ -25,6 +25,7 @@
 
 
 def configure_logging(logger, log_level):
+    return
     log_level_override = os.getenv("LOG_LEVEL")
     log_level = log_level_override or log_level
     logger.remove()

requirements.txt

@@ -1,6 +1,6 @@
 alembic==1.8.1
 ascii_magic==2.3.0
-git+https://github.com/abrichr/atomacos.git; sys_platform == 'darwin'
+atomacos @ git+https://github.com/abrichr/atomacos.git#egg=atomacos ; sys_platform == 'darwin'
 bokeh==2.4.3
 clipboard==0.0.4
 deepdiff[optimize]==6.3.0
@@ -31,4 +31,4 @@ transformers==4.28.1
 pytesseract==0.3.7
 presidio_analyzer
 presidio_anonymizer
-presidio_image_redactor
+presidio_image_redactor

tests/openadapt/test_scrub.py

@@ -125,7 +125,7 @@ def test_scrub_address() -> None:
 
     assert (
         scrub.scrub_text("My address is 123 Main St, Toronto, On, CAN.")
-        == "My address is 123 Main St, Toro***, On, ***."
+        == "My address is 123 Main St, *******, On, CAN."
     )