__init__.py 9.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251
  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. 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. from searx.utils import detect_language
  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. def replace_auto_language(search_query: SearchQuery):
  35. """
  36. Do nothing except if `search_query.lang` is "auto".
  37. In this case:
  38. * the value "auto" is replaced by the detected language of the query.
  39. The default value is "all" when no language is detected.
  40. * `search_query.locale` is updated accordingly
  41. Use :py:obj:`searx.utils.detect_language` with `only_search_languages=True` to keep
  42. only languages supported by the engines.
  43. """
  44. if search_query.lang != 'auto':
  45. return
  46. detected_lang = detect_language(search_query.query, threshold=0.3, only_search_languages=True)
  47. if detected_lang is None:
  48. # fallback to 'all' if no language has been detected
  49. search_query.lang = 'all'
  50. search_query.locale = None
  51. return
  52. search_query.lang = detected_lang
  53. try:
  54. search_query.locale = babel.Locale.parse(search_query.lang)
  55. except babel.core.UnknownLocaleError:
  56. search_query.locale = None
  57. class Search:
  58. """Search information container"""
  59. __slots__ = "search_query", "result_container", "start_time", "actual_timeout"
  60. def __init__(self, search_query: SearchQuery):
  61. """Initialize the Search
  62. search_query is copied
  63. """
  64. # init vars
  65. super().__init__()
  66. self.result_container = ResultContainer()
  67. self.start_time = None
  68. self.actual_timeout = None
  69. self.search_query = copy(search_query)
  70. self.update_search_query(self.search_query)
  71. def update_search_query(self, search_query: SearchQuery):
  72. """Update search_query.
  73. call replace_auto_language to replace the "auto" language
  74. """
  75. replace_auto_language(search_query)
  76. def search_external_bang(self):
  77. """
  78. Check if there is a external bang.
  79. If yes, update self.result_container and return True
  80. """
  81. if self.search_query.external_bang:
  82. self.result_container.redirect_url = get_bang_url(self.search_query)
  83. # This means there was a valid bang and the
  84. # rest of the search does not need to be continued
  85. if isinstance(self.result_container.redirect_url, str):
  86. return True
  87. return False
  88. def search_answerers(self):
  89. """
  90. Check if an answer return a result.
  91. If yes, update self.result_container and return True
  92. """
  93. answerers_results = ask(self.search_query)
  94. if answerers_results:
  95. for results in answerers_results:
  96. self.result_container.extend('answer', results)
  97. return True
  98. return False
  99. # do search-request
  100. def _get_requests(self):
  101. # init vars
  102. requests = []
  103. # max of all selected engine timeout
  104. default_timeout = 0
  105. # start search-reqest for all selected engines
  106. for engineref in self.search_query.engineref_list:
  107. processor = PROCESSORS[engineref.name]
  108. # stop the request now if the engine is suspend
  109. if processor.extend_container_if_suspended(self.result_container):
  110. continue
  111. # set default request parameters
  112. request_params = processor.get_params(self.search_query, engineref.category)
  113. if request_params is None:
  114. continue
  115. counter_inc('engine', engineref.name, 'search', 'count', 'sent')
  116. # append request to list
  117. requests.append((engineref.name, self.search_query.query, request_params))
  118. # update default_timeout
  119. default_timeout = max(default_timeout, processor.engine.timeout)
  120. # adjust timeout
  121. max_request_timeout = settings['outgoing']['max_request_timeout']
  122. actual_timeout = default_timeout
  123. query_timeout = self.search_query.timeout_limit
  124. if max_request_timeout is None and query_timeout is None:
  125. # No max, no user query: default_timeout
  126. pass
  127. elif max_request_timeout is None and query_timeout is not None:
  128. # No max, but user query: From user query except if above default
  129. actual_timeout = min(default_timeout, query_timeout)
  130. elif max_request_timeout is not None and query_timeout is None:
  131. # Max, no user query: Default except if above max
  132. actual_timeout = min(default_timeout, max_request_timeout)
  133. elif max_request_timeout is not None and query_timeout is not None:
  134. # Max & user query: From user query except if above max
  135. actual_timeout = min(query_timeout, max_request_timeout)
  136. logger.debug(
  137. "actual_timeout={0} (default_timeout={1}, ?timeout_limit={2}, max_request_timeout={3})".format(
  138. actual_timeout, default_timeout, query_timeout, max_request_timeout
  139. )
  140. )
  141. return requests, actual_timeout
  142. def search_multiple_requests(self, requests):
  143. # pylint: disable=protected-access
  144. search_id = str(uuid4())
  145. for engine_name, query, request_params in requests:
  146. th = threading.Thread( # pylint: disable=invalid-name
  147. target=PROCESSORS[engine_name].search,
  148. args=(query, request_params, self.result_container, self.start_time, self.actual_timeout),
  149. name=search_id,
  150. )
  151. th._timeout = False
  152. th._engine_name = engine_name
  153. th.start()
  154. for th in threading.enumerate(): # pylint: disable=invalid-name
  155. if th.name == search_id:
  156. remaining_time = max(0.0, self.actual_timeout - (default_timer() - self.start_time))
  157. th.join(remaining_time)
  158. if th.is_alive():
  159. th._timeout = True
  160. self.result_container.add_unresponsive_engine(th._engine_name, 'timeout')
  161. PROCESSORS[th._engine_name].logger.error('engine timeout')
  162. def search_standard(self):
  163. """
  164. Update self.result_container, self.actual_timeout
  165. """
  166. requests, self.actual_timeout = self._get_requests()
  167. # send all search-request
  168. if requests:
  169. self.search_multiple_requests(requests)
  170. # return results, suggestions, answers and infoboxes
  171. return True
  172. # do search-request
  173. def search(self) -> ResultContainer:
  174. self.start_time = default_timer()
  175. if not self.search_external_bang():
  176. if not self.search_answerers():
  177. self.search_standard()
  178. return self.result_container
  179. class SearchWithPlugins(Search):
  180. """Inherit from the Search class, add calls to the plugins."""
  181. __slots__ = 'ordered_plugin_list', 'request'
  182. def __init__(self, search_query: SearchQuery, ordered_plugin_list, request: flask.Request):
  183. super().__init__(search_query)
  184. self.ordered_plugin_list = ordered_plugin_list
  185. self.result_container.on_result = self._on_result
  186. # pylint: disable=line-too-long
  187. # get the "real" request to use it outside the Flask context.
  188. # see
  189. # * https://github.com/pallets/flask/blob/d01d26e5210e3ee4cbbdef12f05c886e08e92852/src/flask/globals.py#L55
  190. # * https://github.com/pallets/werkzeug/blob/3c5d3c9bd0d9ce64590f0af8997a38f3823b368d/src/werkzeug/local.py#L548-L559
  191. # * https://werkzeug.palletsprojects.com/en/2.0.x/local/#werkzeug.local.LocalProxy._get_current_object
  192. # pylint: enable=line-too-long
  193. self.request = request._get_current_object()
  194. def _on_result(self, result):
  195. return plugins.call(self.ordered_plugin_list, 'on_result', self.request, self, result)
  196. def search(self) -> ResultContainer:
  197. if plugins.call(self.ordered_plugin_list, 'pre_search', self.request, self):
  198. super().search()
  199. plugins.call(self.ordered_plugin_list, 'post_search', self.request, self)
  200. self.result_container.close()
  201. return self.result_container