add test; move function to util

schematic/__init__.py

@@ -40,6 +40,10 @@
 from schematic.loader import LOADER
 from schematic.version import __version__
 from dotenv import load_dotenv
+from schematic.utils.remove_sensitive_data_utils import (
+    redact_string,
+    redacted_sensitive_data_in_exception,
+)
 
 Synapse.allow_client_caching(False)
 logger = logging.getLogger(__name__)
@@ -58,59 +62,6 @@
 USER_AGENT |= USER_AGENT_LIBRARY
 
 
-class FilterSensitiveData:
-    """A custom span processor that filters out sensitive data from the spans.
-    It filters out the data from the attributes and events of the spans.
-
-    Args:
-        SpanProcessor (opentelemetry.sdk.trace.SpanProcessor): The base class that provides hooks for processing spans during their lifecycle
-    """
-
-    def __init__(self) -> None:
-        self.sensitive_patterns = {
-            "google_sheets": r"https://sheets\.googleapis\.com/v4/spreadsheets/[\w-]+"
-        }
-
-        self._compiled_patterns = {
-            name: re.compile(pattern)
-            for name, pattern in self.sensitive_patterns.items()
-        }
-
-    def _redact_string(self, value: str) -> str:
-        """remove sensitive data from a string
-
-        Args:
-            value (str): a string that may contain sensitive data
-
-        Returns:
-            str: remove sensitive data from string
-        """
-        redacted = value
-        for pattern_name, pattern in self._compiled_patterns.items():
-            redacted = pattern.sub(f"[REDACTED_{pattern_name.upper()}]", redacted)
-        return redacted
-
-    def redacted_sensitive_data_in_exception(
-        self, exception_attributes: Dict[str, str]
-    ) -> Dict[str, str]:
-        """remove sensitive data in exception
-
-        Args:
-            exception_attributes (dict):a dictionary of exception attributes
-
-        Returns:
-            dict: a dictionary of exception attributes with sensitive data redacted
-        """
-        redacted_exception_attributes = {}
-        for key, value in exception_attributes.items():
-            # remove sensitive information from exception message and stacktrace
-            if key == "exception.message" or key == "exception.stacktrace":
-                redacted_exception_attributes[key] = self._redact_string(value)
-            else:
-                redacted_exception_attributes[key] = value
-        return redacted_exception_attributes
-
-
 class AttributePropagatingSpanProcessor(SpanProcessor):
     """A custom span processor that propagates specific attributes from the parent span
     to the child span when the child span is started.
@@ -277,11 +228,9 @@ def _readable_span_alternate(self: SpanSdk) -> ReadableSpan:
     redacted_events = []
     for event in self._events:
         attributes = event.attributes
-        redacted_event_attributes = (
-            FilterSensitiveData().redacted_sensitive_data_in_exception(attributes)
-        )
+        redacted_event_attributes = redacted_sensitive_data_in_exception(attributes)
         redacted_event = Event(
-            name=FilterSensitiveData()._redact_string(event.name),
+            name=redact_string(event.name),
             attributes=redacted_event_attributes,
             timestamp=event.timestamp,
         )

schematic/utils/remove_sensitive_data_utils.py

@@ -0,0 +1,44 @@
+from typing import Dict
+import re
+
+
+def redact_string(value: str) -> str:
+    """remove sensitive data from a string
+
+    Args:
+        value (str): a string that may contain sensitive data
+
+    Returns:
+        str: remove sensitive data from string
+    """
+    sensitive_patterns = {
+        "google_sheets": r"https://sheets\.googleapis\.com/v4/spreadsheets/[\w-]+"
+    }
+    _compiled_patterns = {
+        name: re.compile(pattern) for name, pattern in sensitive_patterns.items()
+    }
+    redacted = value
+    for pattern_name, pattern in _compiled_patterns.items():
+        redacted = pattern.sub(f"[REDACTED_{pattern_name.upper()}]", redacted)
+    return redacted
+
+
+def redacted_sensitive_data_in_exception(
+    exception_attributes: Dict[str, str]
+) -> Dict[str, str]:
+    """remove sensitive data in exception
+
+    Args:
+        exception_attributes (dict):a dictionary of exception attributes
+
+    Returns:
+        dict: a dictionary of exception attributes with sensitive data redacted
+    """
+    redacted_exception_attributes = {}
+    for key, value in exception_attributes.items():
+        # remove sensitive information from exception message and stacktrace
+        if key == "exception.message" or key == "exception.stacktrace":
+            redacted_exception_attributes[key] = redact_string(value)
+        else:
+            redacted_exception_attributes[key] = value
+    return redacted_exception_attributes

tests/unit/test_filter_sensitive_data.py

@@ -0,0 +1,33 @@
+from schematic.utils.remove_sensitive_data_utils import (
+    redact_string,
+    redacted_sensitive_data_in_exception,
+)
+
+
+class TestFilterSensitiveData:
+    def test_redact_string(self) -> None:
+        # given a string with sensitive data, make sure that they are redacted
+        sensitive_data = "googleapiclient.errors.HttpError: <HttpError 400 when requesting https://sheets.googleapis.com/v4/spreadsheets/11234budyhf:batchUpdate?fields=%2A&alt=json returned abc>"
+        redacted_data = redact_string(sensitive_data)
+        assert (
+            redacted_data
+            == "googleapiclient.errors.HttpError: <HttpError 400 when requesting [REDACTED_GOOGLE_SHEETS]:batchUpdate?fields=%2A&alt=json returned abc>"
+        )
+
+    def test_redacted_sensitive_data_in_exception(self) -> None:
+        # given a dictionary of exception attributes, make sure that sensitive data is redacted
+        exception_attributes = {
+            "exception.message": "googleapiclient.errors.HttpError: <HttpError 400 when requesting https://sheets.googleapis.com/v4/spreadsheets/11234budyhf:batchUpdate?fields=%2A&alt=json returned>",
+            "exception.stacktrace": 'Traceback (most recent call last):\n  File "<stdin>", line 1, in <module>\n  File "<string>", line 1, in <module>\n  File "/usr/local/lib/python3.7/dist-packages/googleapiclient/_helpers.py", line 134, in positional_wrapper\n    return wrapped(*args, **kwargs)\n  File "/usr/local/lib/python3.7/dist-packages/googleapiclient/http.py", line 905, in execute\n    raise HttpError(resp, content, uri=self.uri)\ngoogleapiclient.errors.HttpError: <HttpError 400 when requesting https://sheets.googleapis.com/v4/spreadsheets/11234budyhf:batchUpdate?fields=%2A&alt=json returned>',
+        }
+        redacted_exception_attributes = redacted_sensitive_data_in_exception(
+            exception_attributes
+        )
+        assert (
+            redacted_exception_attributes["exception.message"]
+            == "googleapiclient.errors.HttpError: <HttpError 400 when requesting [REDACTED_GOOGLE_SHEETS]:batchUpdate?fields=%2A&alt=json returned>"
+        )
+        assert (
+            redacted_exception_attributes["exception.stacktrace"]
+            == 'Traceback (most recent call last):\n  File "<stdin>", line 1, in <module>\n  File "<string>", line 1, in <module>\n  File "/usr/local/lib/python3.7/dist-packages/googleapiclient/_helpers.py", line 134, in positional_wrapper\n    return wrapped(*args, **kwargs)\n  File "/usr/local/lib/python3.7/dist-packages/googleapiclient/http.py", line 905, in execute\n    raise HttpError(resp, content, uri=self.uri)\ngoogleapiclient.errors.HttpError: <HttpError 400 when requesting [REDACTED_GOOGLE_SHEETS]:batchUpdate?fields=%2A&alt=json returned>'
+        )