forked from zaclys/searxng
161 lines
6.2 KiB
Python
161 lines
6.2 KiB
Python
from searx.exceptions import SearxParameterException
|
|
from searx.query import RawTextQuery, VALID_LANGUAGE_CODE
|
|
from searx.engines import categories, engines
|
|
from searx.search import SearchQuery, EngineRef
|
|
|
|
|
|
# remove duplicate queries.
|
|
# FIXME: does not fix "!music !soundcloud", because the categories are 'none' and 'music'
|
|
def deduplicate_engineref_list(engineref_list):
|
|
engineref_dict = {q.category + '|' + q.name: q for q in engineref_list}
|
|
return engineref_dict.values()
|
|
|
|
|
|
def get_search_query_from_webapp(preferences, form):
|
|
# no text for the query ?
|
|
if not form.get('q'):
|
|
raise SearxParameterException('q', '')
|
|
|
|
# set blocked engines
|
|
disabled_engines = preferences.engines.get_disabled()
|
|
|
|
# parse query, if tags are set, which change
|
|
# the serch engine or search-language
|
|
raw_text_query = RawTextQuery(form['q'], disabled_engines)
|
|
|
|
# set query
|
|
query = raw_text_query.getQuery()
|
|
|
|
# get and check page number
|
|
pageno_param = form.get('pageno', '1')
|
|
if not pageno_param.isdigit() or int(pageno_param) < 1:
|
|
raise SearxParameterException('pageno', pageno_param)
|
|
query_pageno = int(pageno_param)
|
|
|
|
# get language
|
|
# set specific language if set on request, query or preferences
|
|
# TODO support search with multible languages
|
|
if len(raw_text_query.languages):
|
|
query_lang = raw_text_query.languages[-1]
|
|
elif 'language' in form:
|
|
query_lang = form.get('language')
|
|
else:
|
|
query_lang = preferences.get_value('language')
|
|
|
|
# check language
|
|
if not VALID_LANGUAGE_CODE.match(query_lang):
|
|
raise SearxParameterException('language', query_lang)
|
|
|
|
# get safesearch
|
|
if 'safesearch' in form:
|
|
query_safesearch = form.get('safesearch')
|
|
# first check safesearch
|
|
if not query_safesearch.isdigit():
|
|
raise SearxParameterException('safesearch', query_safesearch)
|
|
query_safesearch = int(query_safesearch)
|
|
else:
|
|
query_safesearch = preferences.get_value('safesearch')
|
|
|
|
# safesearch : second check
|
|
if query_safesearch < 0 or query_safesearch > 2:
|
|
raise SearxParameterException('safesearch', query_safesearch)
|
|
|
|
# get time_range
|
|
query_time_range = form.get('time_range')
|
|
|
|
# check time_range
|
|
if query_time_range not in ('None', None, '', 'day', 'week', 'month', 'year'):
|
|
raise SearxParameterException('time_range', query_time_range)
|
|
|
|
# query_engines
|
|
query_engineref_list = raw_text_query.enginerefs
|
|
|
|
# timeout_limit
|
|
query_timeout = raw_text_query.timeout_limit
|
|
if query_timeout is None and 'timeout_limit' in form:
|
|
raw_time_limit = form.get('timeout_limit')
|
|
if raw_time_limit in ['None', '']:
|
|
raw_time_limit = None
|
|
else:
|
|
try:
|
|
query_timeout = float(raw_time_limit)
|
|
except ValueError:
|
|
raise SearxParameterException('timeout_limit', raw_time_limit)
|
|
|
|
# query_categories
|
|
query_categories = []
|
|
|
|
# if engines are calculated from query,
|
|
# set categories by using that informations
|
|
if query_engineref_list and raw_text_query.specific:
|
|
additional_categories = set()
|
|
for engineref in query_engineref_list:
|
|
if engineref.from_bang:
|
|
additional_categories.add('none')
|
|
else:
|
|
additional_categories.add(engineref.category)
|
|
query_categories = list(additional_categories)
|
|
|
|
# otherwise, using defined categories to
|
|
# calculate which engines should be used
|
|
else:
|
|
# set categories/engines
|
|
load_default_categories = True
|
|
for pd_name, pd in form.items():
|
|
if pd_name == 'categories':
|
|
query_categories.extend(categ for categ in map(str.strip, pd.split(',')) if categ in categories)
|
|
elif pd_name == 'engines':
|
|
pd_engines = [EngineRef(engineref, engines[engineref].categories[0])
|
|
for engine in map(str.strip, pd.split(',')) if engine in engines]
|
|
if pd_engines:
|
|
query_engineref_list.extend(pd_engines)
|
|
load_default_categories = False
|
|
elif pd_name.startswith('category_'):
|
|
category = pd_name[9:]
|
|
|
|
# if category is not found in list, skip
|
|
if category not in categories:
|
|
continue
|
|
|
|
if pd != 'off':
|
|
# add category to list
|
|
query_categories.append(category)
|
|
elif category in query_categories:
|
|
# remove category from list if property is set to 'off'
|
|
query_categories.remove(category)
|
|
|
|
if not load_default_categories:
|
|
if not query_categories:
|
|
query_categories = list(set(engine['category']
|
|
for engine in query_engineref_list))
|
|
else:
|
|
# if no category is specified for this search,
|
|
# using user-defined default-configuration which
|
|
# (is stored in cookie)
|
|
if not query_categories:
|
|
cookie_categories = preferences.get_value('categories')
|
|
for ccateg in cookie_categories:
|
|
if ccateg in categories:
|
|
query_categories.append(ccateg)
|
|
|
|
# if still no category is specified, using general
|
|
# as default-category
|
|
if not query_categories:
|
|
query_categories = ['general']
|
|
|
|
# using all engines for that search, which are
|
|
# declared under the specific categories
|
|
for categ in query_categories:
|
|
query_engineref_list.extend(EngineRef(engine.name, categ)
|
|
for engine in categories[categ]
|
|
if (engine.name, categ) not in disabled_engines)
|
|
|
|
query_engineref_list = deduplicate_engineref_list(query_engineref_list)
|
|
external_bang = raw_text_query.external_bang
|
|
|
|
return (SearchQuery(query, query_engineref_list, query_categories,
|
|
query_lang, query_safesearch, query_pageno,
|
|
query_time_range, query_timeout, preferences,
|
|
external_bang=external_bang),
|
|
raw_text_query)
|