@@ -20,21 +20,15 @@ import gc
2020import threading
2121from time import time
2222from uuid import uuid4
23−from urllib.parse import urlparse
2423from _thread import start_new_thread
2524
26−import requests.exceptions
27−import searx.poolrequests as requests_lib
28−from searx.engines import engines, settings
25+from searx import settings
2926from searx.answerers import ask
3027from searx.external_bang import get_bang_url
31−from searx.utils import gen_useragent
3228from searx.results import ResultContainer
3329from searx import logger
3430from searx.plugins import plugins
35−from searx.exceptions import (SearxEngineAccessDeniedException, SearxEngineCaptchaException,
36− SearxEngineTooManyRequestsException,)
37−from searx.metrology.error_recorder import record_exception, record_error
31+from searx.search.processors import processors, initialize as initialize_processors
3832
3933
4034logger = logger.getChild('search')
@@ -51,6 +45,11 @@ else:
5145 sys.exit(1)
5246
5347
48+def initialize(settings_engines=None):
49+ settings_engines = settings_engines or settings['engines']
50+ initialize_processors(settings_engines)
51+
52+
5453class EngineRef:
5554
5655 __slots__ = 'name', 'category', 'from_bang'
@@ -110,231 +109,6 @@ class SearchQuery:
110109 and self.external_bang == other.external_bang
111110
112111
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−# get default reqest parameter
325−def default_request_params():
326− return {
327− 'method': 'GET',
328− 'headers': {},
329− 'data': {},
330− 'url': '',
331− 'cookies': {},
332− 'verify': True,
333− 'auth': None,
334− 'raise_for_httperror': True
335− }
336−
337−
338112class Search:
339113 """Search information container"""
340114
@@ -375,69 +149,20 @@ class Search:
375149 return True
376150 return False
377151
378− def _is_accepted(self, engine_name, engine):
379− # skip suspended engines
380− if engine.suspend_end_time >= time():
381− logger.debug('Engine currently suspended: %s', engine_name)
382− return False
383−
384− # if paging is not supported, skip
385− if self.search_query.pageno > 1 and not engine.paging:
386− return False
387−
388− # if time_range is not supported, skip
389− if self.search_query.time_range and not engine.time_range_support:
390− return False
391−
392− return True
393−
394− def _get_params(self, engineref, user_agent):
395− if engineref.name not in engines:
396− return None, None
397−
398− engine = engines[engineref.name]
399−
400− if not self._is_accepted(engineref.name, engine):
401− return None, None
402−
403− # set default request parameters
404− request_params = {}
405− if not engine.offline:
406− request_params = default_request_params()
407− request_params['headers']['User-Agent'] = user_agent
408−
409− if hasattr(engine, 'language') and engine.language:
410− request_params['language'] = engine.language
411− else:
412− request_params['language'] = self.search_query.lang
413−
414− request_params['safesearch'] = self.search_query.safesearch
415− request_params['time_range'] = self.search_query.time_range
416−
417− request_params['category'] = engineref.category
418− request_params['pageno'] = self.search_query.pageno
419−
420− with threading.RLock():
421− engine.stats['sent_search_count'] += 1
422−
423− return request_params, engine.timeout
424−
425152 # do search-request
426153 def _get_requests(self):
427154 # init vars
428155 requests = []
429156
430− # set default useragent
431− # user_agent = request.headers.get('User-Agent', '')
432− user_agent = gen_useragent()
433−
434157 # max of all selected engine timeout
435158 default_timeout = 0
436159
437160 # start search-reqest for all selected engines
438161 for engineref in self.search_query.engineref_list:
162+ processor = processors[engineref.name]
163+
439164 # set default request parameters
440− request_params, engine_timeout = self._get_params(engineref, user_agent)
165+ request_params = processor.get_params(self.search_query, engineref.category)
441166 if request_params is None:
442167 continue
443168
@@ -445,7 +170,7 @@ class Search:
445170 requests.append((engineref.name, self.search_query.query, request_params))
446171
447172 # update default_timeout
448− default_timeout = max(default_timeout, engine_timeout)
173+ default_timeout = max(default_timeout, processor.engine.timeout)
449174
450175 # adjust timeout
451176 actual_timeout = default_timeout
@@ -474,8 +199,8 @@ class Search:
474199
475200 for engine_name, query, request_params in requests:
476201 th = threading.Thread(
477− target=search_one_request_safe,
478− args=(engine_name, query, request_params, self.result_container, self.start_time, self.actual_timeout),
202+ target=processors[engine_name].search,
203+ args=(query, request_params, self.result_container, self.start_time, self.actual_timeout),
479204 name=search_id,
480205 )
481206 th._timeout = False