error_recorder.py 5.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142
  1. import typing
  2. import inspect
  3. import logging
  4. from json import JSONDecodeError
  5. from urllib.parse import urlparse
  6. from requests.exceptions import RequestException
  7. from searx.exceptions import SearxXPathSyntaxException, SearxEngineXPathException
  8. from searx import logger
  9. logging.basicConfig(level=logging.INFO)
  10. errors_per_engines = {}
  11. class ErrorContext:
  12. __slots__ = 'filename', 'function', 'line_no', 'code', 'exception_classname', 'log_message', 'log_parameters'
  13. def __init__(self, filename, function, line_no, code, exception_classname, log_message, log_parameters):
  14. self.filename = filename
  15. self.function = function
  16. self.line_no = line_no
  17. self.code = code
  18. self.exception_classname = exception_classname
  19. self.log_message = log_message
  20. self.log_parameters = log_parameters
  21. def __eq__(self, o) -> bool:
  22. if not isinstance(o, ErrorContext):
  23. return False
  24. return self.filename == o.filename and self.function == o.function and self.line_no == o.line_no\
  25. and self.code == o.code and self.exception_classname == o.exception_classname\
  26. and self.log_message == o.log_message and self.log_parameters == o.log_parameters
  27. def __hash__(self):
  28. return hash((self.filename, self.function, self.line_no, self.code, self.exception_classname, self.log_message,
  29. self.log_parameters))
  30. def __repr__(self):
  31. return "ErrorContext({!r}, {!r}, {!r}, {!r}, {!r}, {!r})".\
  32. format(self.filename, self.line_no, self.code, self.exception_classname, self.log_message,
  33. self.log_parameters)
  34. def add_error_context(engine_name: str, error_context: ErrorContext) -> None:
  35. errors_for_engine = errors_per_engines.setdefault(engine_name, {})
  36. errors_for_engine[error_context] = errors_for_engine.get(error_context, 0) + 1
  37. logger.debug('⚠️ %s: %s', engine_name, str(error_context))
  38. def get_trace(traces):
  39. previous_trace = traces[-1]
  40. for trace in reversed(traces):
  41. if trace.filename.endswith('searx/search.py'):
  42. if previous_trace.filename.endswith('searx/poolrequests.py'):
  43. return trace
  44. if previous_trace.filename.endswith('requests/models.py'):
  45. return trace
  46. return previous_trace
  47. previous_trace = trace
  48. return traces[-1]
  49. def get_hostname(exc: RequestException) -> typing.Optional[None]:
  50. url = exc.request.url
  51. if url is None and exc.response is not None:
  52. url = exc.response.url
  53. return urlparse(url).netloc
  54. def get_request_exception_messages(exc: RequestException)\
  55. -> typing.Tuple[typing.Optional[str], typing.Optional[str], typing.Optional[str]]:
  56. url = None
  57. status_code = None
  58. reason = None
  59. hostname = None
  60. if exc.request is not None:
  61. url = exc.request.url
  62. if url is None and exc.response is not None:
  63. url = exc.response.url
  64. if url is not None:
  65. hostname = str(urlparse(url).netloc)
  66. if exc.response is not None:
  67. status_code = str(exc.response.status_code)
  68. reason = exc.response.reason
  69. return (status_code, reason, hostname)
  70. def get_messages(exc, filename) -> typing.Tuple:
  71. if isinstance(exc, JSONDecodeError):
  72. return (exc.msg, )
  73. if isinstance(exc, TypeError):
  74. return (str(exc), )
  75. if isinstance(exc, ValueError) and 'lxml' in filename:
  76. return (str(exc), )
  77. if isinstance(exc, RequestException):
  78. return get_request_exception_messages(exc)
  79. if isinstance(exc, SearxXPathSyntaxException):
  80. return (exc.xpath_str, exc.message)
  81. if isinstance(exc, SearxEngineXPathException):
  82. return (exc.xpath_str, exc.message)
  83. return ()
  84. def get_exception_classname(exc: Exception) -> str:
  85. exc_class = exc.__class__
  86. exc_name = exc_class.__qualname__
  87. exc_module = exc_class.__module__
  88. if exc_module is None or exc_module == str.__class__.__module__:
  89. return exc_name
  90. return exc_module + '.' + exc_name
  91. def get_error_context(framerecords, exception_classname, log_message, log_parameters) -> ErrorContext:
  92. searx_frame = get_trace(framerecords)
  93. filename = searx_frame.filename
  94. function = searx_frame.function
  95. line_no = searx_frame.lineno
  96. code = searx_frame.code_context[0].strip()
  97. del framerecords
  98. return ErrorContext(filename, function, line_no, code, exception_classname, log_message, log_parameters)
  99. def record_exception(engine_name: str, exc: Exception) -> None:
  100. framerecords = inspect.trace()
  101. try:
  102. exception_classname = get_exception_classname(exc)
  103. log_parameters = get_messages(exc, framerecords[-1][1])
  104. error_context = get_error_context(framerecords, exception_classname, None, log_parameters)
  105. add_error_context(engine_name, error_context)
  106. finally:
  107. del framerecords
  108. def record_error(engine_name: str, log_message: str, log_parameters: typing.Optional[typing.Tuple] = None) -> None:
  109. framerecords = list(reversed(inspect.stack()[1:]))
  110. try:
  111. error_context = get_error_context(framerecords, None, log_message, log_parameters or ())
  112. add_error_context(engine_name, error_context)
  113. finally:
  114. del framerecords