Feature/add masking by keys

logredactor/redacting_filter.py

@@ -4,34 +4,35 @@
 
 class RedactingFilter(logging.Filter):
     # Do not try and redact the built in values. With the wrong regex it can break the logging
-    ignore_keys = [
+    ignore_keys = {
         'name', 'levelname', 'levelno', 'pathname', 'filename', 'module',
         'exc_info', 'exc_text', 'stack_info', 'lineno', 'funcName', 'created',
         'msecs', 'relativeCreated', 'thread', 'threadName', 'process',
         'processName', 'args',
-    ]
+    }
 
-    def __init__(self, patterns, default_mask='****'):
+    def __init__(self, patterns, default_mask='****', mask_keys=None):
         super(RedactingFilter, self).__init__()
         self._patterns = patterns
         self._default_mask = str(default_mask)
+        self._mask_keys = set(mask_keys or {})
 
     def filter(self, record):
         d = vars(record)
         for k, content in d.items():
             if k not in self.ignore_keys:
-                d[k] = self.redact(content)
+                d[k] = self.redact(content, k)
 
         # Also clean any contents in args
         if isinstance(record.args, dict):
             for k in record.args.keys():
-                record.args[k] = self.redact(record.args[k])
+                record.args[k] = self.redact(record.args[k], k)
         else:
             record.args = tuple(self.redact(arg) for arg in record.args)
 
         return True
 
-    def redact(self, content):
+    def redact(self, content, key=None):
         if content:
             if isinstance(content, dict):
                 for k, v in content.items():
@@ -40,6 +41,9 @@ def redact(self, content):
             elif isinstance(content, (list, tuple)):
                 for i, v in enumerate(content):
                     content[i] = self.redact(v)
+
+            elif key and key in self._mask_keys:
+                    content = self._default_mask
 
             else:
                 content = isinstance(content, str) and content or str(content)

tests/test_redacting_filter.py

@@ -12,7 +12,8 @@ def get_logger(filters):
         logger.addFilter(
             logredactor.RedactingFilter(
                 filters,
-                default_mask='****'
+                default_mask='****',
+                mask_keys={'phonenumber',}
             )
         )
         return logger
@@ -38,6 +39,12 @@ def test_arg_dict(caplog, logger_setup):
     assert caplog.records[0].message == "foo ****"
 
 
+def test_arg_dict_with_key_to_remove(caplog, logger_setup):
+    logger = logger_setup([re.compile(r'\d{3}')])
+    logger.warning("foo %s", {'phonenumber': '123'})
+    assert caplog.records[0].message == "foo {'phonenumber': '****'}"
+
+
 def test_extra_string_value(caplog, logger_setup):
     logger = logger_setup([re.compile(r'\d{3}')])
     logger.warning("foo", extra={'bar': '123 too'})
@@ -92,3 +99,9 @@ def test_match_group(caplog, logger_setup):
     logger = logger_setup([re.compile(r'(?<=api_key=)[\w-]+')])
     logger.warning("example.com?api_key=this-is-my-key&sort=price")
     assert caplog.records[0].message == "example.com?api_key=****&sort=price"
+
+
+def test_extra_do_redact_specific_key(caplog, logger_setup):
+    logger = logger_setup([re.compile(r'\d{3}')])
+    logger.warning("foo", extra={'phonenumber': 'foobar'})
+    assert caplog.records[0].phonenumber == "****"