requested changes

openadapt/config.py

@@ -40,11 +40,12 @@
     "SCRUB_ENABLED": True,
     "SCRUB_CHAR": "*",
     "SCRUB_LANGUAGE": "en",
+    # TODO support lists in getenv_fallback
+    "SCRUB_FILL_COLOR": (255, 0, 0),
     "SCRUB_CONFIG_TRF": {
         "nlp_engine_name": "spacy",
         "models": [{"lang_code": "en", "model_name": "en_core_web_trf"}],
     },
-    "DEFAULT_SCRUB_FILL_COLOR": (255, 0, 0),
     "SCRUB_IGNORE_ENTITIES": [
         # 'US_PASSPORT',
         # 'US_DRIVER_LICENSE',
@@ -106,3 +107,23 @@ def getenv_fallback(var_name):
     for key, val in locals().items():
         if not key.startswith("_") and key.isupper():
             logger.info(f"{key}={val}")
+
+
+def filter_log_messages(data):
+    """
+    This function filters log messages by ignoring any message that contains a specific string.
+
+    Args:
+      data: The input parameter "data" is expected to be data from a loguru logger.
+
+    Returns:
+      a boolean value indicating whether the message in the input data should be ignored or not. If the
+    message contains any of the messages in the `messages_to_ignore` list, the function returns `False`
+    indicating that the message should be ignored. Otherwise, it returns `True` indicating that the
+    message should not be ignored.
+    """
+    # TODO: ultimately, we want to fix the underlying issues, but for now, we can ignore these messages
+    messages_to_ignore = [
+        "Cannot pickle Objective-C objects",
+    ]
+    return not any(msg in data["message"] for msg in messages_to_ignore)

openadapt/scrub.py

@@ -101,7 +101,9 @@ def scrub_text_all(text: str) -> str:
     return config.SCRUB_CHAR * len(text)
 
 
-def scrub_image(image: Image, fill_color=config.DEFAULT_SCRUB_FILL_COLOR) -> Image:
+def scrub_image(
+    image: Image, fill_color=config.SCRUB_FILL_COLOR
+) -> Image:
     """
     Scrub the image of all PII/PHI using Presidio Image Redactor
 
@@ -157,7 +159,9 @@ def _is_scrubbed(old_text: str, new_text: str) -> bool:
     return old_text != new_text
 
 
-def _scrub_text_item(value: str, key: Any, force_scrub_children: bool = False) -> str:
+def _scrub_text_item(
+    value: str, key: Any, force_scrub_children: bool = False
+) -> str:
     """
     Scrubs the value of a dict item.
 
@@ -188,7 +192,11 @@ def _should_scrub_list_item(item: Any, key: Any, list_keys: List[str]) -> bool:
         bool: True if the key and value should be scrubbed, False otherwise
     """
 
-    return isinstance(item, (str, dict)) and isinstance(key, str) and key in list_keys
+    return (
+        isinstance(item, (str, dict))
+        and isinstance(key, str)
+        and key in list_keys
+    )
 
 
 def _scrub_list_item(
@@ -209,7 +217,9 @@ def _scrub_list_item(
         dict/str: The scrubbed dict/value respectively
     """
     if isinstance(item, dict):
-        return scrub_dict(item, list_keys, force_scrub_children=force_scrub_children)
+        return scrub_dict(
+            item, list_keys, force_scrub_children=force_scrub_children
+        )
     return _scrub_text_item(item, key)
 
 
@@ -234,8 +244,12 @@ def scrub_dict(
     scrubbed_dict = {}
     for key, value in input_dict.items():
         if _should_scrub_text(key, value, list_keys, scrub_all):
-            scrubbed_text = _scrub_text_item(value, key, force_scrub_children)
-            if key in ("text", "canonical_text") and _is_scrubbed(value, scrubbed_text):
+            scrubbed_text = _scrub_text_item(
+                value, key, force_scrub_children
+            )
+            if key in ("text", "canonical_text") and _is_scrubbed(
+                value, scrubbed_text
+            ):
                 force_scrub_children = True
             scrubbed_dict[key] = scrubbed_text
         elif isinstance(value, list):

openadapt/utils.py

@@ -29,18 +29,11 @@ def configure_logging(logger, log_level):
     logger.add(
         sys.stderr,
         level=log_level,
-        filter=filter_log_messages if config.IGNORE_WARNINGS else None,
+        filter=config.filter_log_messages if config.IGNORE_WARNINGS else None,
     )
     logger.debug(f"{log_level=}")
 
 
-def filter_log_messages(data):
-    messages_to_ignore = [
-        "Cannot pickle Objective-C objects",
-    ]
-    return not any(msg in data["message"] for msg in messages_to_ignore)
-
-
 def row2dict(row, follow=True):
     if isinstance(row, dict):
         return row
@@ -173,6 +166,7 @@ def get_monitor_dims():
     return monitor_width, monitor_height
 
 
+# TODO: move parameters to config
 def draw_ellipse(
     x,
     y,

openadapt/visualize.py

@@ -157,9 +157,9 @@ def main():
         event_dicts = scrub.scrub_list_dicts(event_dicts)
     logger.info(f"event_dicts=\n{pformat(event_dicts)}")
 
-    r2d_recording = row2dict(recording)
+    recording_dict = row2dict(recording)
     if SCRUB:
-        r2d_recording = scrub.scrub_dict(r2d_recording)
+        recording_dict = scrub.scrub_dict(recording_dict)
 
     rows = [
         row(
@@ -169,7 +169,7 @@ def main():
         ),
         row(
             Div(
-                text=f"{dict2html(r2d_recording)}",
+                text=f"{dict2html(recording_dict)}",
             ),
         ),
         row(
@@ -197,13 +197,13 @@ def main():
         mask_utf8 = image2utf8(mask)
         width, height = image.size
 
-        r2d_ae = row2dict(action_event)
-        r2d_ae_we = row2dict(action_event.window_event)
-        
+        action_event_dict = row2dict(action_event)
+        window_event_dict = row2dict(action_event.window_event)
+
         if SCRUB:
-            r2d_ae = scrub.scrub_dict(r2d_ae)
-            r2d_ae_we = scrub.scrub_dict(r2d_ae_we)
-            
+            action_event_dict = scrub.scrub_dict(action_event_dict)
+            window_event_dict = scrub.scrub_dict(window_event_dict)
+
         rows.append(
             [
                 row(
@@ -230,14 +230,14 @@ def main():
                             >
                         </div>
                         <table>
-                            {dict2html(r2d_ae_we , None)}
+                            {dict2html(window_event_dict , None)}
                         </table>
                     """,
                     ),
                     Div(
                         text=f"""
                         <table>
-                            {dict2html(r2d_ae)}
+                            {dict2html(action_event_dict)}
                         </table>
                     """
                     ),

tests/openadapt/test_scrub.py

@@ -37,9 +37,7 @@ def test_scrub_image() -> None:
 
     # Count the number of pixels having the color of the mask
     mask_pixels = sum(
-        1
-        for pixel in scrubbed_image.getdata()
-        if pixel == config.DEFAULT_SCRUB_FILL_COLOR
+        1 for pixel in scrubbed_image.getdata() if pixel == config.SCRUB_FILL_COLOR
     )
     total_pixels = scrubbed_image.width * scrubbed_image.height
 
@@ -104,9 +102,7 @@ def test_scrub_credit_card() -> None:
     """
 
     assert (
-        scrub.scrub_text(
-            "My credit card number is 4234-5678-9012-3456 and "
-        )
+        scrub.scrub_text("My credit card number is 4234-5678-9012-3456 and ")
     ) == "My credit card number is ******************* and "
 
 
@@ -219,5 +215,6 @@ def test_scrub_all_together() -> None:
         " He was born on 01/01/1980."
     )
 
+
 if __name__ == "__main__":
     test_scrub_image()