searxng/searx/search.py

346 lines
12 KiB
Python
Raw Normal View History

'''
searx is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
searx is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with searx. If not, see < http://www.gnu.org/licenses/ >.
(C) 2013- by Adam Tauber, <asciimoo@gmail.com>
'''
import threading
import searx.poolrequests as requests_lib
2014-12-14 00:18:01 +00:00
from time import time
from searx import settings
from searx.engines import (
2014-10-19 10:41:04 +00:00
categories, engines
)
from searx.languages import language_codes
from searx.utils import gen_useragent
from searx.query import Query
from searx.results import ResultContainer
2015-01-09 03:13:05 +00:00
from searx import logger
2014-07-07 11:59:27 +00:00
2015-01-09 03:13:05 +00:00
logger = logger.getChild('search')
2014-07-07 11:59:27 +00:00
number_of_searches = 0
def search_request_wrapper(fn, url, engine_name, **kwargs):
ret = None
engine = engines[engine_name]
try:
ret = fn(url, **kwargs)
with threading.RLock():
engine.continuous_errors = 0
engine.suspend_end_time = 0
2015-01-09 03:30:55 +00:00
except:
# increase errors stats
with threading.RLock():
engine.stats['errors'] += 1
engine.continuous_errors += 1
engine.suspend_end_time = time() + min(60, engine.continuous_errors)
# print engine name and specific error message
2015-01-09 03:30:55 +00:00
logger.exception('engine crash: {0}'.format(engine_name))
return ret
def threaded_requests(requests):
2014-12-14 00:18:01 +00:00
timeout_limit = max(r[2]['timeout'] for r in requests)
search_start = time()
2014-12-19 12:59:41 +00:00
for fn, url, request_args, engine_name in requests:
2014-12-18 09:11:56 +00:00
request_args['timeout'] = timeout_limit
th = threading.Thread(
target=search_request_wrapper,
args=(fn, url, engine_name),
2014-12-08 22:55:11 +00:00
kwargs=request_args,
name='search_request',
)
2014-12-19 12:59:41 +00:00
th._engine_name = engine_name
th.start()
for th in threading.enumerate():
2014-12-08 22:55:11 +00:00
if th.name == 'search_request':
2014-12-14 00:18:01 +00:00
remaining_time = max(0.0, timeout_limit - (time() - search_start))
th.join(remaining_time)
if th.isAlive():
2015-01-09 03:13:05 +00:00
logger.warning('engine timeout: {0}'.format(th._engine_name))
2014-12-14 00:18:01 +00:00
# get default reqest parameter
2014-07-07 11:59:27 +00:00
def default_request_params():
return {
'method': 'GET',
'headers': {},
'data': {},
'url': '',
'cookies': {},
'verify': True
}
2014-07-07 11:59:27 +00:00
# create a callback wrapper for the search engine results
def make_callback(engine_name, callback, params, result_container):
2014-07-07 11:59:27 +00:00
# creating a callback wrapper for the search engine results
def process_callback(response, **kwargs):
2015-02-13 10:17:00 +00:00
# check if redirect comparing to the True value,
# because resp can be a Mock object, and any attribut name returns something.
if response.is_redirect is True:
2015-02-13 12:02:37 +00:00
logger.debug('{0} redirect on: {1}'.format(engine_name, response))
2015-02-13 10:17:00 +00:00
return
2014-07-07 11:59:27 +00:00
response.search_params = params
search_duration = time() - params['started']
# update stats with current page-load-time
with threading.RLock():
engines[engine_name].stats['page_load_time'] += search_duration
timeout_overhead = 0.2 # seconds
timeout_limit = engines[engine_name].timeout + timeout_overhead
if search_duration > timeout_limit:
with threading.RLock():
engines[engine_name].stats['errors'] += 1
return
2014-09-28 14:51:41 +00:00
# callback
search_results = callback(response)
2014-09-28 14:51:41 +00:00
# add results
2014-07-07 11:59:27 +00:00
for result in search_results:
result['engine'] = engine_name
result_container.extend(engine_name, search_results)
2014-09-28 14:51:41 +00:00
2014-07-07 11:59:27 +00:00
return process_callback
2014-09-28 14:51:41 +00:00
class Search(object):
"""Search information container"""
def __init__(self, request):
# init vars
super(Search, self).__init__()
self.query = None
self.engines = []
self.categories = []
self.paging = False
self.pageno = 1
self.lang = 'all'
# set blocked engines
self.blocked_engines = request.preferences.engines.get_disabled()
self.result_container = ResultContainer()
self.request_data = {}
# set specific language if set
self.lang = request.preferences.get_value('language')
# set request method
if request.method == 'POST':
self.request_data = request.form
else:
self.request_data = request.args
# TODO better exceptions
if not self.request_data.get('q'):
raise Exception('noquery')
# set pagenumber
pageno_param = self.request_data.get('pageno', '1')
if not pageno_param.isdigit() or int(pageno_param) < 1:
pageno_param = 1
self.pageno = int(pageno_param)
2014-10-19 10:41:04 +00:00
# parse query, if tags are set, which change
# the serch engine or search-language
query_obj = Query(self.request_data['q'], self.blocked_engines)
query_obj.parse_query()
# set query
self.query = query_obj.getSearchQuery()
# get last selected language in query, if possible
# TODO support search with multible languages
if len(query_obj.languages):
self.lang = query_obj.languages[-1]
self.engines = query_obj.engines
self.categories = []
2014-10-19 10:41:04 +00:00
# if engines are calculated from query,
# set categories by using that informations
if self.engines and query_obj.specific:
self.categories = list(set(engine['category']
2014-02-07 01:45:12 +00:00
for engine in self.engines))
2014-10-19 10:41:04 +00:00
# otherwise, using defined categories to
# calculate which engines should be used
else:
# set categories/engines
load_default_categories = True
for pd_name, pd in self.request_data.items():
if pd_name == 'categories':
2015-09-01 13:14:02 +00:00
self.categories.extend(categ for categ in map(unicode.strip, pd.split(',')) if categ in categories)
elif pd_name == 'engines':
pd_engines = [{'category': engines[engine].categories[0],
'name': engine}
2015-09-01 13:14:02 +00:00
for engine in map(unicode.strip, pd.split(',')) if engine in engines]
if pd_engines:
self.engines.extend(pd_engines)
load_default_categories = False
elif pd_name.startswith('category_'):
category = pd_name[9:]
# if category is not found in list, skip
2014-10-19 10:41:04 +00:00
if category not in categories:
continue
if pd != 'off':
# add category to list
self.categories.append(category)
elif category in self.categories:
# remove category from list if property is set to 'off'
self.categories.remove(category)
if not load_default_categories:
if not self.categories:
self.categories = list(set(engine['category']
for engine in self.engines))
return
2014-10-19 10:41:04 +00:00
# if no category is specified for this search,
# using user-defined default-configuration which
# (is stored in cookie)
2014-02-11 12:13:51 +00:00
if not self.categories:
cookie_categories = request.cookies.get('categories', '')
cookie_categories = cookie_categories.split(',')
for ccateg in cookie_categories:
if ccateg in categories:
self.categories.append(ccateg)
2014-10-19 10:41:04 +00:00
# if still no category is specified, using general
# as default-category
2014-02-11 12:13:51 +00:00
if not self.categories:
self.categories = ['general']
2014-10-19 10:41:04 +00:00
# using all engines for that search, which are
# declared under the specific categories
for categ in self.categories:
self.engines.extend({'category': categ,
'name': engine.name}
for engine in categories[categ]
if (engine.name, categ) not in self.blocked_engines)
2014-02-09 00:07:18 +00:00
# remove suspended engines
self.engines = [e for e in self.engines
if engines[e['name']].suspend_end_time <= time()]
# do search-request
2014-07-07 11:59:27 +00:00
def search(self, request):
global number_of_searches
# init vars
2014-07-07 11:59:27 +00:00
requests = []
# increase number of searches
2014-07-07 11:59:27 +00:00
number_of_searches += 1
# set default useragent
2014-10-19 10:41:04 +00:00
# user_agent = request.headers.get('User-Agent', '')
2014-07-07 11:59:27 +00:00
user_agent = gen_useragent()
# start search-reqest for all selected engines
2014-07-07 11:59:27 +00:00
for selected_engine in self.engines:
if selected_engine['name'] not in engines:
continue
engine = engines[selected_engine['name']]
# if paging is not supported, skip
2014-07-07 11:59:27 +00:00
if self.pageno > 1 and not engine.paging:
continue
2014-10-19 10:41:04 +00:00
# if search-language is set and engine does not
# provide language-support, skip
2014-07-07 11:59:27 +00:00
if self.lang != 'all' and not engine.language_support:
continue
# set default request parameters
2014-07-07 11:59:27 +00:00
request_params = default_request_params()
request_params['headers']['User-Agent'] = user_agent
request_params['category'] = selected_engine['category']
2014-12-18 09:11:56 +00:00
request_params['started'] = time()
2014-07-07 11:59:27 +00:00
request_params['pageno'] = self.pageno
if hasattr(engine, 'language') and engine.language:
request_params['language'] = engine.language
else:
request_params['language'] = self.lang
# 0 = None, 1 = Moderate, 2 = Strict
request_params['safesearch'] = request.preferences.get_value('safesearch')
2014-10-19 10:41:04 +00:00
# update request parameters dependent on
# search-engine (contained in engines folder)
2014-12-18 09:11:56 +00:00
engine.request(self.query.encode('utf-8'), request_params)
2014-07-07 11:59:27 +00:00
if request_params['url'] is None:
# TODO add support of offline engines
pass
# create a callback wrapper for the search engine results
2014-07-07 11:59:27 +00:00
callback = make_callback(
selected_engine['name'],
engine.response,
request_params,
self.result_container)
2014-07-07 11:59:27 +00:00
2014-10-19 10:41:04 +00:00
# create dictionary which contain all
# informations about the request
2014-07-07 11:59:27 +00:00
request_args = dict(
headers=request_params['headers'],
hooks=dict(response=callback),
cookies=request_params['cookies'],
timeout=engine.timeout,
verify=request_params['verify']
2014-07-07 11:59:27 +00:00
)
# specific type of request (GET or POST)
2014-07-07 11:59:27 +00:00
if request_params['method'] == 'GET':
req = requests_lib.get
2014-07-07 11:59:27 +00:00
else:
req = requests_lib.post
2014-07-07 11:59:27 +00:00
request_args['data'] = request_params['data']
# ignoring empty urls
if not request_params['url']:
continue
# append request to list
requests.append((req, request_params['url'],
request_args,
selected_engine['name']))
2014-12-19 18:40:40 +00:00
if not requests:
2015-03-08 22:30:31 +00:00
return self
# send all search-request
threaded_requests(requests)
2014-09-28 14:51:41 +00:00
# return results, suggestions, answers and infoboxes
2015-03-08 22:30:31 +00:00
return self