background.py 5.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171
  1. # SPDX-License-Identifier: AGPL-3.0-or-later
  2. # lint: pylint
  3. # pylint: disable=missing-module-docstring
  4. # pyright: basic
  5. import json
  6. import time
  7. import threading
  8. import os
  9. import signal
  10. from typing import Dict, Union, List, Any, Tuple, Optional
  11. from typing_extensions import TypedDict, Literal
  12. import redis.exceptions
  13. from searx import logger, settings, searx_debug
  14. from searx.redisdb import client as get_redis_client
  15. from searx.exceptions import SearxSettingsException
  16. from searx.search.processors import PROCESSORS
  17. from searx.search.checker import Checker
  18. from searx.search.checker.scheduler import scheduler_function
  19. REDIS_RESULT_KEY = 'SearXNG_checker_result'
  20. REDIS_LOCK_KEY = 'SearXNG_checker_lock'
  21. CheckerResult = Union['CheckerOk', 'CheckerErr', 'CheckerOther']
  22. class CheckerOk(TypedDict):
  23. """Checking the engines succeeded"""
  24. status: Literal['ok']
  25. engines: Dict[str, 'EngineResult']
  26. timestamp: int
  27. class CheckerErr(TypedDict):
  28. """Checking the engines failed"""
  29. status: Literal['error']
  30. timestamp: int
  31. class CheckerOther(TypedDict):
  32. """The status is unknown or disabled"""
  33. status: Literal['unknown', 'disabled']
  34. EngineResult = Union['EngineOk', 'EngineErr']
  35. class EngineOk(TypedDict):
  36. """Checking the engine succeeded"""
  37. success: Literal[True]
  38. class EngineErr(TypedDict):
  39. """Checking the engine failed"""
  40. success: Literal[False]
  41. errors: Dict[str, List[str]]
  42. def _get_interval(every: Any, error_msg: str) -> Tuple[int, int]:
  43. if isinstance(every, int):
  44. return (every, every)
  45. if (
  46. not isinstance(every, (tuple, list))
  47. or len(every) != 2 # type: ignore
  48. or not isinstance(every[0], int)
  49. or not isinstance(every[1], int)
  50. ):
  51. raise SearxSettingsException(error_msg, None)
  52. return (every[0], every[1])
  53. def get_result() -> CheckerResult:
  54. client = get_redis_client()
  55. if client is None:
  56. # without Redis, the checker is disabled
  57. return {'status': 'disabled'}
  58. serialized_result: Optional[bytes] = client.get(REDIS_RESULT_KEY)
  59. if serialized_result is None:
  60. # the Redis key does not exist
  61. return {'status': 'unknown'}
  62. return json.loads(serialized_result)
  63. def _set_result(result: CheckerResult):
  64. client = get_redis_client()
  65. if client is None:
  66. # without Redis, the function does nothing
  67. return
  68. client.set(REDIS_RESULT_KEY, json.dumps(result))
  69. def _timestamp():
  70. return int(time.time() / 3600) * 3600
  71. def run():
  72. try:
  73. # use a Redis lock to make sure there is no checker running at the same time
  74. # (this should not happen, this is a safety measure)
  75. with get_redis_client().lock(REDIS_LOCK_KEY, blocking_timeout=60, timeout=3600):
  76. logger.info('Starting checker')
  77. result: CheckerOk = {'status': 'ok', 'engines': {}, 'timestamp': _timestamp()}
  78. for name, processor in PROCESSORS.items():
  79. logger.debug('Checking %s engine', name)
  80. checker = Checker(processor)
  81. checker.run()
  82. if checker.test_results.successful:
  83. result['engines'][name] = {'success': True}
  84. else:
  85. result['engines'][name] = {'success': False, 'errors': checker.test_results.errors}
  86. _set_result(result)
  87. logger.info('Check done')
  88. except redis.exceptions.LockError:
  89. _set_result({'status': 'error', 'timestamp': _timestamp()})
  90. logger.exception('Error while running the checker')
  91. except Exception: # pylint: disable=broad-except
  92. _set_result({'status': 'error', 'timestamp': _timestamp()})
  93. logger.exception('Error while running the checker')
  94. def _signal_handler(_signum: int, _frame: Any):
  95. t = threading.Thread(target=run)
  96. t.daemon = True
  97. t.start()
  98. def initialize():
  99. if hasattr(signal, 'SIGUSR1'):
  100. # Windows doesn't support SIGUSR1
  101. logger.info('Send SIGUSR1 signal to pid %i to start the checker', os.getpid())
  102. signal.signal(signal.SIGUSR1, _signal_handler)
  103. # special case when debug is activate
  104. if searx_debug and settings['checker']['off_when_debug']:
  105. logger.info('debug mode: checker is disabled')
  106. return
  107. # check value of checker.scheduling.every now
  108. scheduling = settings['checker']['scheduling']
  109. if scheduling is None or not scheduling:
  110. logger.info('Checker scheduler is disabled')
  111. return
  112. # make sure there is a Redis connection
  113. if get_redis_client() is None:
  114. logger.error('The checker requires Redis')
  115. return
  116. # start the background scheduler
  117. every_range = _get_interval(scheduling.get('every', (300, 1800)), 'checker.scheduling.every is not a int or list')
  118. start_after_range = _get_interval(
  119. scheduling.get('start_after', (300, 1800)), 'checker.scheduling.start_after is not a int or list'
  120. )
  121. t = threading.Thread(
  122. target=scheduler_function,
  123. args=(start_after_range[0], start_after_range[1], every_range[0], every_range[1], run),
  124. name='checker_scheduler',
  125. )
  126. t.daemon = True
  127. t.start()