__init__.py 8.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214
  1. # SPDX-License-Identifier: AGPL-3.0-or-later
  2. # pylint: disable=missing-module-docstring, too-few-public-methods
  3. import threading
  4. from copy import copy
  5. from timeit import default_timer
  6. from uuid import uuid4
  7. import flask
  8. from flask import copy_current_request_context
  9. import babel
  10. from searx import settings
  11. from searx.answerers import ask
  12. from searx.external_bang import get_bang_url
  13. from searx.results import ResultContainer
  14. from searx import logger
  15. from searx.plugins import plugins
  16. from searx.search.models import EngineRef, SearchQuery
  17. from searx.engines import load_engines
  18. from searx.network import initialize as initialize_network, check_network_configuration
  19. from searx.metrics import initialize as initialize_metrics, counter_inc, histogram_observe_time
  20. from searx.search.processors import PROCESSORS, initialize as initialize_processors
  21. from searx.search.checker import initialize as initialize_checker
  22. logger = logger.getChild('search')
  23. def initialize(settings_engines=None, enable_checker=False, check_network=False, enable_metrics=True):
  24. settings_engines = settings_engines or settings['engines']
  25. load_engines(settings_engines)
  26. initialize_network(settings_engines, settings['outgoing'])
  27. if check_network:
  28. check_network_configuration()
  29. initialize_metrics([engine['name'] for engine in settings_engines], enable_metrics)
  30. initialize_processors(settings_engines)
  31. if enable_checker:
  32. initialize_checker()
  33. class Search:
  34. """Search information container"""
  35. __slots__ = "search_query", "result_container", "start_time", "actual_timeout"
  36. def __init__(self, search_query: SearchQuery):
  37. """Initialize the Search"""
  38. # init vars
  39. super().__init__()
  40. self.search_query = search_query
  41. self.result_container = ResultContainer()
  42. self.start_time = None
  43. self.actual_timeout = None
  44. def search_external_bang(self):
  45. """
  46. Check if there is a external bang.
  47. If yes, update self.result_container and return True
  48. """
  49. if self.search_query.external_bang:
  50. self.result_container.redirect_url = get_bang_url(self.search_query)
  51. # This means there was a valid bang and the
  52. # rest of the search does not need to be continued
  53. if isinstance(self.result_container.redirect_url, str):
  54. return True
  55. return False
  56. def search_answerers(self):
  57. """
  58. Check if an answer return a result.
  59. If yes, update self.result_container and return True
  60. """
  61. answerers_results = ask(self.search_query)
  62. if answerers_results:
  63. for results in answerers_results:
  64. self.result_container.extend('answer', results)
  65. return True
  66. return False
  67. # do search-request
  68. def _get_requests(self):
  69. # init vars
  70. requests = []
  71. # max of all selected engine timeout
  72. default_timeout = 0
  73. # start search-request for all selected engines
  74. for engineref in self.search_query.engineref_list:
  75. processor = PROCESSORS[engineref.name]
  76. # stop the request now if the engine is suspend
  77. if processor.extend_container_if_suspended(self.result_container):
  78. continue
  79. # set default request parameters
  80. request_params = processor.get_params(self.search_query, engineref.category)
  81. if request_params is None:
  82. continue
  83. counter_inc('engine', engineref.name, 'search', 'count', 'sent')
  84. # append request to list
  85. requests.append((engineref.name, self.search_query.query, request_params))
  86. # update default_timeout
  87. default_timeout = max(default_timeout, processor.engine.timeout)
  88. # adjust timeout
  89. max_request_timeout = settings['outgoing']['max_request_timeout']
  90. actual_timeout = default_timeout
  91. query_timeout = self.search_query.timeout_limit
  92. if max_request_timeout is None and query_timeout is None:
  93. # No max, no user query: default_timeout
  94. pass
  95. elif max_request_timeout is None and query_timeout is not None:
  96. # No max, but user query: From user query except if above default
  97. actual_timeout = min(default_timeout, query_timeout)
  98. elif max_request_timeout is not None and query_timeout is None:
  99. # Max, no user query: Default except if above max
  100. actual_timeout = min(default_timeout, max_request_timeout)
  101. elif max_request_timeout is not None and query_timeout is not None:
  102. # Max & user query: From user query except if above max
  103. actual_timeout = min(query_timeout, max_request_timeout)
  104. logger.debug(
  105. "actual_timeout={0} (default_timeout={1}, ?timeout_limit={2}, max_request_timeout={3})".format(
  106. actual_timeout, default_timeout, query_timeout, max_request_timeout
  107. )
  108. )
  109. return requests, actual_timeout
  110. def search_multiple_requests(self, requests):
  111. # pylint: disable=protected-access
  112. search_id = str(uuid4())
  113. for engine_name, query, request_params in requests:
  114. _search = copy_current_request_context(PROCESSORS[engine_name].search)
  115. th = threading.Thread( # pylint: disable=invalid-name
  116. target=_search,
  117. args=(query, request_params, self.result_container, self.start_time, self.actual_timeout),
  118. name=search_id,
  119. )
  120. th._timeout = False
  121. th._engine_name = engine_name
  122. th.start()
  123. for th in threading.enumerate(): # pylint: disable=invalid-name
  124. if th.name == search_id:
  125. remaining_time = max(0.0, self.actual_timeout - (default_timer() - self.start_time))
  126. th.join(remaining_time)
  127. if th.is_alive():
  128. th._timeout = True
  129. self.result_container.add_unresponsive_engine(th._engine_name, 'timeout')
  130. PROCESSORS[th._engine_name].logger.error('engine timeout')
  131. def search_standard(self):
  132. """
  133. Update self.result_container, self.actual_timeout
  134. """
  135. requests, self.actual_timeout = self._get_requests()
  136. # send all search-request
  137. if requests:
  138. self.search_multiple_requests(requests)
  139. # return results, suggestions, answers and infoboxes
  140. return True
  141. # do search-request
  142. def search(self) -> ResultContainer:
  143. self.start_time = default_timer()
  144. if not self.search_external_bang():
  145. if not self.search_answerers():
  146. self.search_standard()
  147. return self.result_container
  148. class SearchWithPlugins(Search):
  149. """Inherit from the Search class, add calls to the plugins."""
  150. __slots__ = 'ordered_plugin_list', 'request'
  151. def __init__(self, search_query: SearchQuery, ordered_plugin_list, request: flask.Request):
  152. super().__init__(search_query)
  153. self.ordered_plugin_list = ordered_plugin_list
  154. self.result_container.on_result = self._on_result
  155. # pylint: disable=line-too-long
  156. # get the "real" request to use it outside the Flask context.
  157. # see
  158. # * https://github.com/pallets/flask/blob/d01d26e5210e3ee4cbbdef12f05c886e08e92852/src/flask/globals.py#L55
  159. # * https://github.com/pallets/werkzeug/blob/3c5d3c9bd0d9ce64590f0af8997a38f3823b368d/src/werkzeug/local.py#L548-L559
  160. # * https://werkzeug.palletsprojects.com/en/2.0.x/local/#werkzeug.local.LocalProxy._get_current_object
  161. # pylint: enable=line-too-long
  162. self.request = request._get_current_object()
  163. def _on_result(self, result):
  164. return plugins.call(self.ordered_plugin_list, 'on_result', self.request, self, result)
  165. def search(self) -> ResultContainer:
  166. if plugins.call(self.ordered_plugin_list, 'pre_search', self.request, self):
  167. super().search()
  168. plugins.call(self.ordered_plugin_list, 'post_search', self.request, self)
  169. self.result_container.close()
  170. return self.result_container