@@ -1,541 +0,0 @@
1−'''
2−searx is free software: you can redistribute it and/or modify
3−it under the terms of the GNU Affero General Public License as published by
4−the Free Software Foundation, either version 3 of the License, or
5−(at your option) any later version.
6−
7−searx is distributed in the hope that it will be useful,
8−but WITHOUT ANY WARRANTY; without even the implied warranty of
9−MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
10−GNU Affero General Public License for more details.
11−
12−You should have received a copy of the GNU Affero General Public License
13−along with searx. If not, see < http://www.gnu.org/licenses/ >.
14−
15−(C) 2013- by Adam Tauber, <asciimoo@gmail.com>
16−'''
17−
18−import typing
19−import gc
20−import threading
21−from time import time
22−from uuid import uuid4
23−from urllib.parse import urlparse
24−from _thread import start_new_thread
25−
26−import requests.exceptions
27−import searx.poolrequests as requests_lib
28−from searx.engines import engines, settings
29−from searx.answerers import ask
30−from searx.external_bang import get_bang_url
31−from searx.utils import gen_useragent
32−from searx.results import ResultContainer
33−from searx import logger
34−from searx.plugins import plugins
35−from searx.exceptions import (SearxEngineAccessDeniedException, SearxEngineCaptchaException,
36− SearxEngineTooManyRequestsException,)
37−from searx.metrology.error_recorder import record_exception, record_error
38−
39−
40−logger = logger.getChild('search')
41−
42−max_request_timeout = settings.get('outgoing', {}).get('max_request_timeout' or None)
43−if max_request_timeout is None:
44− logger.info('max_request_timeout={0}'.format(max_request_timeout))
45−else:
46− if isinstance(max_request_timeout, float):
47− logger.info('max_request_timeout={0} second(s)'.format(max_request_timeout))
48− else:
49− logger.critical('outgoing.max_request_timeout if defined has to be float')
50− import sys
51− sys.exit(1)
52−
53−
54−class EngineRef:
55−
56− __slots__ = 'name', 'category', 'from_bang'
57−
58− def __init__(self, name: str, category: str, from_bang: bool=False):
59− self.name = name
60− self.category = category
61− self.from_bang = from_bang
62−
63− def __repr__(self):
64− return "EngineRef({!r}, {!r}, {!r})".format(self.name, self.category, self.from_bang)
65−
66− def __eq__(self, other):
67− return self.name == other.name and self.category == other.category and self.from_bang == other.from_bang
68−
69−
70−class SearchQuery:
71− """container for all the search parameters (query, language, etc...)"""
72−
73− __slots__ = 'query', 'engineref_list', 'categories', 'lang', 'safesearch', 'pageno', 'time_range',\
74− 'timeout_limit', 'external_bang'
75−
76− def __init__(self,
77− query: str,
78− engineref_list: typing.List[EngineRef],
79− categories: typing.List[str],
80− lang: str,
81− safesearch: int,
82− pageno: int,
83− time_range: typing.Optional[str],
84− timeout_limit: typing.Optional[float]=None,
85− external_bang: typing.Optional[str]=None):
86− self.query = query
87− self.engineref_list = engineref_list
88− self.categories = categories
89− self.lang = lang
90− self.safesearch = safesearch
91− self.pageno = pageno
92− self.time_range = time_range
93− self.timeout_limit = timeout_limit
94− self.external_bang = external_bang
95−
96− def __repr__(self):
97− return "SearchQuery({!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r})".\
98− format(self.query, self.engineref_list, self.categories, self.lang, self.safesearch,
99− self.pageno, self.time_range, self.timeout_limit, self.external_bang)
100−
101− def __eq__(self, other):
102− return self.query == other.query\
103− and self.engineref_list == other.engineref_list\
104− and self.categories == self.categories\
105− and self.lang == other.lang\
106− and self.safesearch == other.safesearch\
107− and self.pageno == other.pageno\
108− and self.time_range == other.time_range\
109− and self.timeout_limit == other.timeout_limit\
110− and self.external_bang == other.external_bang
111−
112−
113−def send_http_request(engine, request_params):
114− # create dictionary which contain all
115− # informations about the request
116− request_args = dict(
117− headers=request_params['headers'],
118− cookies=request_params['cookies'],
119− verify=request_params['verify'],
120− auth=request_params['auth']
121− )
122−
123− # setting engine based proxies
124− if hasattr(engine, 'proxies'):
125− request_args['proxies'] = requests_lib.get_proxies(engine.proxies)
126−
127− # max_redirects
128− max_redirects = request_params.get('max_redirects')
129− if max_redirects:
130− request_args['max_redirects'] = max_redirects
131−
132− # soft_max_redirects
133− soft_max_redirects = request_params.get('soft_max_redirects', max_redirects or 0)
134−
135− # raise_for_status
136− request_args['raise_for_httperror'] = request_params.get('raise_for_httperror', False)
137−
138− # specific type of request (GET or POST)
139− if request_params['method'] == 'GET':
140− req = requests_lib.get
141− else:
142− req = requests_lib.post
143−
144− request_args['data'] = request_params['data']
145−
146− # send the request
147− response = req(request_params['url'], **request_args)
148−
149− # check soft limit of the redirect count
150− if len(response.history) > soft_max_redirects:
151− # unexpected redirect : record an error
152− # but the engine might still return valid results.
153− status_code = str(response.status_code or '')
154− reason = response.reason or ''
155− hostname = str(urlparse(response.url or '').netloc)
156− record_error(engine.name,
157− '{} redirects, maximum: {}'.format(len(response.history), soft_max_redirects),
158− (status_code, reason, hostname))
159−
160− return response
161−
162−
163−def search_one_http_request(engine, query, request_params):
164− # update request parameters dependent on
165− # search-engine (contained in engines folder)
166− engine.request(query, request_params)
167−
168− # ignoring empty urls
169− if request_params['url'] is None:
170− return None
171−
172− if not request_params['url']:
173− return None
174−
175− # send request
176− response = send_http_request(engine, request_params)
177−
178− # parse the response
179− response.search_params = request_params
180− return engine.response(response)
181−
182−
183−def search_one_http_request_safe(engine_name, query, request_params, result_container, start_time, timeout_limit):
184− # set timeout for all HTTP requests
185− requests_lib.set_timeout_for_thread(timeout_limit, start_time=start_time)
186− # reset the HTTP total time
187− requests_lib.reset_time_for_thread()
188−
189− #
190− engine = engines[engine_name]
191−
192− # suppose everything will be alright
193− requests_exception = False
194− suspended_time = None
195−
196− try:
197− # send requests and parse the results
198− search_results = search_one_http_request(engine, query, request_params)
199−
200− # check if the engine accepted the request
201− if search_results is not None:
202− # yes, so add results
203− result_container.extend(engine_name, search_results)
204−
205− # update engine time when there is no exception
206− engine_time = time() - start_time
207− page_load_time = requests_lib.get_time_for_thread()
208− result_container.add_timing(engine_name, engine_time, page_load_time)
209− with threading.RLock():
210− engine.stats['engine_time'] += engine_time
211− engine.stats['engine_time_count'] += 1
212− # update stats with the total HTTP time
213− engine.stats['page_load_time'] += page_load_time
214− engine.stats['page_load_count'] += 1
215− except Exception as e:
216− record_exception(engine_name, e)
217−
218− # Timing
219− engine_time = time() - start_time
220− page_load_time = requests_lib.get_time_for_thread()
221− result_container.add_timing(engine_name, engine_time, page_load_time)
222−
223− # Record the errors
224− with threading.RLock():
225− engine.stats['errors'] += 1
226−
227− if (issubclass(e.__class__, requests.exceptions.Timeout)):
228− result_container.add_unresponsive_engine(engine_name, 'HTTP timeout')
229− # requests timeout (connect or read)
230− logger.error("engine {0} : HTTP requests timeout"
231− "(search duration : {1} s, timeout: {2} s) : {3}"
232− .format(engine_name, engine_time, timeout_limit, e.__class__.__name__))
233− requests_exception = True
234− elif (issubclass(e.__class__, requests.exceptions.RequestException)):
235− result_container.add_unresponsive_engine(engine_name, 'HTTP error')
236− # other requests exception
237− logger.exception("engine {0} : requests exception"
238− "(search duration : {1} s, timeout: {2} s) : {3}"
239− .format(engine_name, engine_time, timeout_limit, e))
240− requests_exception = True
241− elif (issubclass(e.__class__, SearxEngineCaptchaException)):
242− result_container.add_unresponsive_engine(engine_name, 'CAPTCHA required')
243− logger.exception('engine {0} : CAPTCHA')
244− suspended_time = e.suspended_time # pylint: disable=no-member
245− elif (issubclass(e.__class__, SearxEngineTooManyRequestsException)):
246− result_container.add_unresponsive_engine(engine_name, 'too many requests')
247− logger.exception('engine {0} : Too many requests')
248− suspended_time = e.suspended_time # pylint: disable=no-member
249− elif (issubclass(e.__class__, SearxEngineAccessDeniedException)):
250− result_container.add_unresponsive_engine(engine_name, 'blocked')
251− logger.exception('engine {0} : Searx is blocked')
252− suspended_time = e.suspended_time # pylint: disable=no-member
253− else:
254− result_container.add_unresponsive_engine(engine_name, 'unexpected crash')
255− # others errors
256− logger.exception('engine {0} : exception : {1}'.format(engine_name, e))
257− else:
258− if getattr(threading.current_thread(), '_timeout', False):
259− record_error(engine_name, 'Timeout')
260−
261− # suspend the engine if there is an HTTP error
262− # or suspended_time is defined
263− with threading.RLock():
264− if requests_exception or suspended_time:
265− # update continuous_errors / suspend_end_time
266− engine.continuous_errors += 1
267− if suspended_time is None:
268− suspended_time = min(settings['search']['max_ban_time_on_fail'],
269− engine.continuous_errors * settings['search']['ban_time_on_fail'])
270− engine.suspend_end_time = time() + suspended_time
271− else:
272− # reset the suspend variables
273− engine.continuous_errors = 0
274− engine.suspend_end_time = 0
275−
276−
277−def record_offline_engine_stats_on_error(engine, result_container, start_time):
278− engine_time = time() - start_time
279− result_container.add_timing(engine.name, engine_time, engine_time)
280−
281− with threading.RLock():
282− engine.stats['errors'] += 1
283−
284−
285−def search_one_offline_request(engine, query, request_params):
286− return engine.search(query, request_params)
287−
288−
289−def search_one_offline_request_safe(engine_name, query, request_params, result_container, start_time, timeout_limit):
290− engine = engines[engine_name]
291−
292− try:
293− search_results = search_one_offline_request(engine, query, request_params)
294−
295− if search_results:
296− result_container.extend(engine_name, search_results)
297−
298− engine_time = time() - start_time
299− result_container.add_timing(engine_name, engine_time, engine_time)
300− with threading.RLock():
301− engine.stats['engine_time'] += engine_time
302− engine.stats['engine_time_count'] += 1
303−
304− except ValueError as e:
305− record_exception(engine_name, e)
306− record_offline_engine_stats_on_error(engine, result_container, start_time)
307− logger.exception('engine {0} : invalid input : {1}'.format(engine_name, e))
308− except Exception as e:
309− record_exception(engine_name, e)
310− record_offline_engine_stats_on_error(engine, result_container, start_time)
311− result_container.add_unresponsive_engine(engine_name, 'unexpected crash', str(e))
312− logger.exception('engine {0} : exception : {1}'.format(engine_name, e))
313− else:
314− if getattr(threading.current_thread(), '_timeout', False):
315− record_error(engine_name, 'Timeout')
316−
317−
318−def search_one_request_safe(engine_name, query, request_params, result_container, start_time, timeout_limit):
319− if engines[engine_name].offline:
320− return search_one_offline_request_safe(engine_name, query, request_params, result_container, start_time, timeout_limit) # noqa
321− return search_one_http_request_safe(engine_name, query, request_params, result_container, start_time, timeout_limit)
322−
323−
324−def search_multiple_requests(requests, result_container, start_time, timeout_limit):
325− search_id = uuid4().__str__()
326−
327− for engine_name, query, request_params in requests:
328− th = threading.Thread(
329− target=search_one_request_safe,
330− args=(engine_name, query, request_params, result_container, start_time, timeout_limit),
331− name=search_id,
332− )
333− th._timeout = False
334− th._engine_name = engine_name
335− th.start()
336−
337− for th in threading.enumerate():
338− if th.name == search_id:
339− remaining_time = max(0.0, timeout_limit - (time() - start_time))
340− th.join(remaining_time)
341− if th.is_alive():
342− th._timeout = True
343− result_container.add_unresponsive_engine(th._engine_name, 'timeout')
344− logger.warning('engine timeout: {0}'.format(th._engine_name))
345−
346−
347−# get default reqest parameter
348−def default_request_params():
349− return {
350− 'method': 'GET',
351− 'headers': {},
352− 'data': {},
353− 'url': '',
354− 'cookies': {},
355− 'verify': True,
356− 'auth': None,
357− 'raise_for_httperror': True
358− }
359−
360−
361−class Search:
362− """Search information container"""
363−
364− __slots__ = "search_query", "result_container", "start_time", "actual_timeout"
365−
366− def __init__(self, search_query):
367− # init vars
368− super().__init__()
369− self.search_query = search_query
370− self.result_container = ResultContainer()
371− self.start_time = None
372− self.actual_timeout = None
373−
374− def search_external_bang(self):
375− """
376− Check if there is a external bang.
377− If yes, update self.result_container and return True
378− """
379− if self.search_query.external_bang:
380− self.result_container.redirect_url = get_bang_url(self.search_query)
381−
382− # This means there was a valid bang and the
383− # rest of the search does not need to be continued
384− if isinstance(self.result_container.redirect_url, str):
385− return True
386− return False
387−
388− def search_answerers(self):
389− """
390− Check if an answer return a result.
391− If yes, update self.result_container and return True
392− """
393− answerers_results = ask(self.search_query)
394−
395− if answerers_results:
396− for results in answerers_results:
397− self.result_container.extend('answer', results)
398− return True
399− return False
400−
401− def _is_accepted(self, engine_name, engine):
402− # skip suspended engines
403− if engine.suspend_end_time >= time():
404− logger.debug('Engine currently suspended: %s', engine_name)
405− return False
406−
407− # if paging is not supported, skip
408− if self.search_query.pageno > 1 and not engine.paging:
409− return False
410−
411− # if time_range is not supported, skip
412− if self.search_query.time_range and not engine.time_range_support:
413− return False
414−
415− return True
416−
417− def _get_params(self, engineref, user_agent):
418− if engineref.name not in engines:
419− return None, None
420−
421− engine = engines[engineref.name]
422−
423− if not self._is_accepted(engineref.name, engine):
424− return None, None
425−
426− # set default request parameters
427− request_params = {}
428− if not engine.offline:
429− request_params = default_request_params()
430− request_params['headers']['User-Agent'] = user_agent
431−
432− if hasattr(engine, 'language') and engine.language:
433− request_params['language'] = engine.language
434− else:
435− request_params['language'] = self.search_query.lang
436−
437− request_params['safesearch'] = self.search_query.safesearch
438− request_params['time_range'] = self.search_query.time_range
439−
440− request_params['category'] = engineref.category
441− request_params['pageno'] = self.search_query.pageno
442−
443− with threading.RLock():
444− engine.stats['sent_search_count'] += 1
445−
446− return request_params, engine.timeout
447−
448− # do search-request
449− def _get_requests(self):
450− # init vars
451− requests = []
452−
453− # set default useragent
454− # user_agent = request.headers.get('User-Agent', '')
455− user_agent = gen_useragent()
456−
457− # max of all selected engine timeout
458− default_timeout = 0
459−
460− # start search-reqest for all selected engines
461− for engineref in self.search_query.engineref_list:
462− # set default request parameters
463− request_params, engine_timeout = self._get_params(engineref, user_agent)
464− if request_params is None:
465− continue
466−
467− # append request to list
468− requests.append((engineref.name, self.search_query.query, request_params))
469−
470− # update default_timeout
471− default_timeout = max(default_timeout, engine_timeout)
472−
473− # adjust timeout
474− actual_timeout = default_timeout
475− query_timeout = self.search_query.timeout_limit
476−
477− if max_request_timeout is None and query_timeout is None:
478− # No max, no user query: default_timeout
479− pass
480− elif max_request_timeout is None and query_timeout is not None:
481− # No max, but user query: From user query except if above default
482− actual_timeout = min(default_timeout, query_timeout)
483− elif max_request_timeout is not None and query_timeout is None:
484− # Max, no user query: Default except if above max
485− actual_timeout = min(default_timeout, max_request_timeout)
486− elif max_request_timeout is not None and query_timeout is not None:
487− # Max & user query: From user query except if above max
488− actual_timeout = min(query_timeout, max_request_timeout)
489−
490− logger.debug("actual_timeout={0} (default_timeout={1}, ?timeout_limit={2}, max_request_timeout={3})"
491− .format(actual_timeout, default_timeout, query_timeout, max_request_timeout))
492−
493− return requests, actual_timeout
494−
495− def search_standard(self):
496− """
497− Update self.result_container, self.actual_timeout
498− """
499− requests, self.actual_timeout = self._get_requests()
500−
501− # send all search-request
502− if requests:
503− search_multiple_requests(requests, self.result_container, self.start_time, self.actual_timeout)
504− start_new_thread(gc.collect, tuple())
505−
506− # return results, suggestions, answers and infoboxes
507− return True
508−
509− # do search-request
510− def search(self):
511− self.start_time = time()
512−
513− if not self.search_external_bang():
514− if not self.search_answerers():
515− self.search_standard()
516−
517− return self.result_container
518−
519−
520−class SearchWithPlugins(Search):
521− """Similar to the Search class but call the plugins."""
522−
523− __slots__ = 'ordered_plugin_list', 'request'
524−
525− def __init__(self, search_query, ordered_plugin_list, request):
526− super().__init__(search_query)
527− self.ordered_plugin_list = ordered_plugin_list
528− self.request = request
529−
530− def search(self):
531− if plugins.call(self.ordered_plugin_list, 'pre_search', self.request, self):
532− super().search()
533−
534− plugins.call(self.ordered_plugin_list, 'post_search', self.request, self)
535−
536− results = self.result_container.get_ordered_results()
537−
538− for result in results:
539− plugins.call(self.ordered_plugin_list, 'on_result', self.request, self, result)
540−
541− return self.result_container