__init__.py 8.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215
  1. # SPDX-License-Identifier: AGPL-3.0-or-later
  2. # lint: pylint
  3. # pylint: disable=missing-module-docstring, too-few-public-methods
  4. import threading
  5. from copy import copy
  6. from timeit import default_timer
  7. from uuid import uuid4
  8. import flask
  9. from flask import copy_current_request_context
  10. import babel
  11. from searx import settings
  12. from searx.answerers import ask
  13. from searx.external_bang import get_bang_url
  14. from searx.results import ResultContainer
  15. from searx import logger
  16. from searx.plugins import plugins
  17. from searx.search.models import EngineRef, SearchQuery
  18. from searx.engines import load_engines
  19. from searx.network import initialize as initialize_network, check_network_configuration
  20. from searx.metrics import initialize as initialize_metrics, counter_inc, histogram_observe_time
  21. from searx.search.processors import PROCESSORS, initialize as initialize_processors
  22. from searx.search.checker import initialize as initialize_checker
  23. logger = logger.getChild('search')
  24. def initialize(settings_engines=None, enable_checker=False, check_network=False, enable_metrics=True):
  25. settings_engines = settings_engines or settings['engines']
  26. load_engines(settings_engines)
  27. initialize_network(settings_engines, settings['outgoing'])
  28. if check_network:
  29. check_network_configuration()
  30. initialize_metrics([engine['name'] for engine in settings_engines], enable_metrics)
  31. initialize_processors(settings_engines)
  32. if enable_checker:
  33. initialize_checker()
  34. class Search:
  35. """Search information container"""
  36. __slots__ = "search_query", "result_container", "start_time", "actual_timeout"
  37. def __init__(self, search_query: SearchQuery):
  38. """Initialize the Search"""
  39. # init vars
  40. super().__init__()
  41. self.search_query = search_query
  42. self.result_container = ResultContainer()
  43. self.start_time = None
  44. self.actual_timeout = None
  45. def search_external_bang(self):
  46. """
  47. Check if there is a external bang.
  48. If yes, update self.result_container and return True
  49. """
  50. if self.search_query.external_bang:
  51. self.result_container.redirect_url = get_bang_url(self.search_query)
  52. # This means there was a valid bang and the
  53. # rest of the search does not need to be continued
  54. if isinstance(self.result_container.redirect_url, str):
  55. return True
  56. return False
  57. def search_answerers(self):
  58. """
  59. Check if an answer return a result.
  60. If yes, update self.result_container and return True
  61. """
  62. answerers_results = ask(self.search_query)
  63. if answerers_results:
  64. for results in answerers_results:
  65. self.result_container.extend('answer', results)
  66. return True
  67. return False
  68. # do search-request
  69. def _get_requests(self):
  70. # init vars
  71. requests = []
  72. # max of all selected engine timeout
  73. default_timeout = 0
  74. # start search-reqest for all selected engines
  75. for engineref in self.search_query.engineref_list:
  76. processor = PROCESSORS[engineref.name]
  77. # stop the request now if the engine is suspend
  78. if processor.extend_container_if_suspended(self.result_container):
  79. continue
  80. # set default request parameters
  81. request_params = processor.get_params(self.search_query, engineref.category)
  82. if request_params is None:
  83. continue
  84. counter_inc('engine', engineref.name, 'search', 'count', 'sent')
  85. # append request to list
  86. requests.append((engineref.name, self.search_query.query, request_params))
  87. # update default_timeout
  88. default_timeout = max(default_timeout, processor.engine.timeout)
  89. # adjust timeout
  90. max_request_timeout = settings['outgoing']['max_request_timeout']
  91. actual_timeout = default_timeout
  92. query_timeout = self.search_query.timeout_limit
  93. if max_request_timeout is None and query_timeout is None:
  94. # No max, no user query: default_timeout
  95. pass
  96. elif max_request_timeout is None and query_timeout is not None:
  97. # No max, but user query: From user query except if above default
  98. actual_timeout = min(default_timeout, query_timeout)
  99. elif max_request_timeout is not None and query_timeout is None:
  100. # Max, no user query: Default except if above max
  101. actual_timeout = min(default_timeout, max_request_timeout)
  102. elif max_request_timeout is not None and query_timeout is not None:
  103. # Max & user query: From user query except if above max
  104. actual_timeout = min(query_timeout, max_request_timeout)
  105. logger.debug(
  106. "actual_timeout={0} (default_timeout={1}, ?timeout_limit={2}, max_request_timeout={3})".format(
  107. actual_timeout, default_timeout, query_timeout, max_request_timeout
  108. )
  109. )
  110. return requests, actual_timeout
  111. def search_multiple_requests(self, requests):
  112. # pylint: disable=protected-access
  113. search_id = str(uuid4())
  114. for engine_name, query, request_params in requests:
  115. _search = copy_current_request_context(PROCESSORS[engine_name].search)
  116. th = threading.Thread( # pylint: disable=invalid-name
  117. target=_search,
  118. args=(query, request_params, self.result_container, self.start_time, self.actual_timeout),
  119. name=search_id,
  120. )
  121. th._timeout = False
  122. th._engine_name = engine_name
  123. th.start()
  124. for th in threading.enumerate(): # pylint: disable=invalid-name
  125. if th.name == search_id:
  126. remaining_time = max(0.0, self.actual_timeout - (default_timer() - self.start_time))
  127. th.join(remaining_time)
  128. if th.is_alive():
  129. th._timeout = True
  130. self.result_container.add_unresponsive_engine(th._engine_name, 'timeout')
  131. PROCESSORS[th._engine_name].logger.error('engine timeout')
  132. def search_standard(self):
  133. """
  134. Update self.result_container, self.actual_timeout
  135. """
  136. requests, self.actual_timeout = self._get_requests()
  137. # send all search-request
  138. if requests:
  139. self.search_multiple_requests(requests)
  140. # return results, suggestions, answers and infoboxes
  141. return True
  142. # do search-request
  143. def search(self) -> ResultContainer:
  144. self.start_time = default_timer()
  145. if not self.search_external_bang():
  146. if not self.search_answerers():
  147. self.search_standard()
  148. return self.result_container
  149. class SearchWithPlugins(Search):
  150. """Inherit from the Search class, add calls to the plugins."""
  151. __slots__ = 'ordered_plugin_list', 'request'
  152. def __init__(self, search_query: SearchQuery, ordered_plugin_list, request: flask.Request):
  153. super().__init__(search_query)
  154. self.ordered_plugin_list = ordered_plugin_list
  155. self.result_container.on_result = self._on_result
  156. # pylint: disable=line-too-long
  157. # get the "real" request to use it outside the Flask context.
  158. # see
  159. # * https://github.com/pallets/flask/blob/d01d26e5210e3ee4cbbdef12f05c886e08e92852/src/flask/globals.py#L55
  160. # * https://github.com/pallets/werkzeug/blob/3c5d3c9bd0d9ce64590f0af8997a38f3823b368d/src/werkzeug/local.py#L548-L559
  161. # * https://werkzeug.palletsprojects.com/en/2.0.x/local/#werkzeug.local.LocalProxy._get_current_object
  162. # pylint: enable=line-too-long
  163. self.request = request._get_current_object()
  164. def _on_result(self, result):
  165. return plugins.call(self.ordered_plugin_list, 'on_result', self.request, self, result)
  166. def search(self) -> ResultContainer:
  167. if plugins.call(self.ordered_plugin_list, 'pre_search', self.request, self):
  168. super().search()
  169. plugins.call(self.ordered_plugin_list, 'post_search', self.request, self)
  170. self.result_container.close()
  171. return self.result_container