Fix sentry scrubber and logs

src/tribler/core/sentry_reporter/sentry_reporter.py

@@ -9,6 +9,7 @@
 from typing import Any, Dict, Optional
 
 import sentry_sdk
+import sentry_sdk.utils
 from faker import Faker
 from sentry_sdk.integrations.logging import LoggingIntegration, ignore_logger
 from sentry_sdk.integrations.threading import ThreadingIntegration
@@ -19,6 +20,25 @@
  get_value
 )
 
+
+def fix_sentry_logger(sentry_logger):
+ # Sentry log requires reconfiguration to be useful. By default, Sentry does not show even error-level log records
+ # until the debug option for Sentry is enabled. That means that unsuccessful attempts to send report are ignored.
+ # Enabling debug for Sentry is not recommended in production, and it sends the debug messages to stderr via
+ # a separate handler that is not in line with how Tribler handles logs. As a solution, Sentry developers recommend
+ # manual logger reconfiguration: https://github.com/getsentry/sentry-python/issues/1191#issuecomment-1023721841
+
+ for f in list(sentry_logger.filters):
+ sentry_logger.removeFilter(f)
+
+ for h in list(sentry_logger.handlers):
+ sentry_logger.removeHandler(h)
+
+ sentry_logger.setLevel(logging.WARNING)
+
+
+fix_sentry_logger(sentry_sdk.utils.logger)
+
 VALUE = 'value'
 TYPE = 'type'
 LAST_CORE_OUTPUT = 'last_core_output'

src/tribler/core/sentry_reporter/sentry_scrubber.py

@@ -174,17 +174,22 @@ def scrub_entity_recursively(self, entity: Union[str, Dict, List, Any], depth=10
  if isinstance(entity, dict):
  result = {}
  for key, value in entity.items():
- if key in self.dict_keys_for_scrub:
+ if key in self.dict_keys_for_scrub and isinstance(value, str):
  value = value.strip()
  fake_value = obfuscate_string(value)
  placeholder = self.create_placeholder(fake_value)
  self.add_sensitive_pair(value, placeholder)
- result[key] = self.scrub_entity_recursively(value, depth)
+ result[key] = placeholder
+ else:
+ result[key] = self.scrub_entity_recursively(value, depth)
  return result
 
  return entity
 
  def add_sensitive_pair(self, text, placeholder):
+ if not (text and text.strip()): # We should not replace empty substrings in the middle of other strings
+ return
+
  if text in self.sensitive_occurrences:
  return
 

src/tribler/core/sentry_reporter/sentry_tools.py

@@ -118,9 +118,6 @@ def obfuscate_string(s: str, part_of_speech: str = 'noun') -> str:
 
  The same random words will be generated for the same given strings.
  """
- if not s:
- return s
-
  faker = Faker(locale='en_US')
  faker.seed_instance(s)
  return faker.word(part_of_speech=part_of_speech)

src/tribler/core/sentry_reporter/tests/test_sentry_scrubber.py

@@ -218,16 +218,16 @@ def test_scrub_event(scrubber):
  }
  assert scrubber.scrub_event(event) == {
  'the very first item': '<highlight>',
- 'server_name': '<protection>',
+ 'server_name': '<kid>',
  CONTEXTS: {
  REPORTER: {
  'any': {
- 'USERNAME': '<father>',
+ 'USERNAME': '<conference>',
  'USERDOMAIN_ROAMINGPROFILE': '<protection>',
  'PATH': '/users/<highlight>/apps',
  'TMP_WIN': 'C:\\Users\\<restaurant>\\AppData\\Local\\Temp',
- 'USERDOMAIN': '<marriage>',
- 'COMPUTERNAME': '<message>',
+ 'USERDOMAIN': '<tune>',
+ 'COMPUTERNAME': '<lady>',
  },
  STACKTRACE: [
  'Traceback (most recent call last):',
@@ -306,15 +306,20 @@ def test_scrub_dict(scrubber):
  assert scrubber.scrub_entity_recursively(None) is None
  assert scrubber.scrub_entity_recursively({}) == {}
 
- given = {'PATH': '/home/username/some/', 'USERDOMAIN': 'UD', 'USERNAME': 'U', 'REPEATED': 'user username UD U'}
+ assert scrubber.scrub_entity_recursively({'key': [1]}) == {'key': [1]} # non-string values should not lead to error
+
+ given = {'PATH': '/home/username/some/', 'USERDOMAIN': 'UD', 'USERNAME': 'U', 'REPEATED': 'user username UD U',
+ 'key': ''}
  assert scrubber.scrub_entity_recursively(given) == {'PATH': '/home/<highlight>/some/',
  'REPEATED': 'user <highlight> <school> <night>',
  'USERDOMAIN': '<school>',
- 'USERNAME': '<night>'}
+ 'USERNAME': '<night>',
+ 'key': '<dress>'}
 
- assert 'username' in scrubber.sensitive_occurrences.keys()
- assert 'UD' in scrubber.sensitive_occurrences.keys()
- assert 'U' in scrubber.sensitive_occurrences.keys()
+ assert 'username' in scrubber.sensitive_occurrences
+ assert 'UD' in scrubber.sensitive_occurrences
+ assert 'U' in scrubber.sensitive_occurrences
+ assert '' not in scrubber.sensitive_occurrences
 
 
 def test_scrub_list(scrubber):

src/tribler/core/sentry_reporter/tests/test_sentry_tools.py

@@ -102,8 +102,7 @@ def test_extract_dict():
 
 
 OBFUSCATED_STRINGS = [
- (None, None),
- ('', ''),
+ ('', 'dress'),
  ('any', 'challenge'),
  ('string', 'quality'),
 ]