mirror of
https://github.com/searxng/searxng
synced 2024-11-09 01:10:26 +00:00
133 lines
4.5 KiB
Python
133 lines
4.5 KiB
Python
#!/usr/bin/env python
|
|
|
|
'''
|
|
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) 2014 by Thomas Pointhuber, <thomas.pointhuber@gmx.at>
|
|
'''
|
|
|
|
from searx.languages import language_codes
|
|
from searx.engines import (
|
|
categories, engines, engine_shortcuts
|
|
)
|
|
import string
|
|
import re
|
|
|
|
|
|
class Query(object):
|
|
"""parse query"""
|
|
|
|
def __init__(self, query, blocked_engines):
|
|
self.query = query
|
|
self.blocked_engines = []
|
|
|
|
if blocked_engines:
|
|
self.blocked_engines = blocked_engines
|
|
|
|
self.query_parts = []
|
|
self.engines = []
|
|
self.languages = []
|
|
self.specific = False
|
|
|
|
# parse query, if tags are set, which
|
|
# change the serch engine or search-language
|
|
def parse_query(self):
|
|
self.query_parts = []
|
|
|
|
# split query, including whitespaces
|
|
raw_query_parts = re.split(r'(\s+)', self.query)
|
|
|
|
parse_next = True
|
|
|
|
for query_part in raw_query_parts:
|
|
if not parse_next:
|
|
self.query_parts[-1] += query_part
|
|
continue
|
|
|
|
parse_next = False
|
|
|
|
# part does only contain spaces, skip
|
|
if query_part.isspace()\
|
|
or query_part == '':
|
|
parse_next = True
|
|
self.query_parts.append(query_part)
|
|
continue
|
|
|
|
# this force a language
|
|
if query_part[0] == ':':
|
|
lang = query_part[1:].lower()
|
|
|
|
# check if any language-code is equal with
|
|
# declared language-codes
|
|
for lc in language_codes:
|
|
lang_id, lang_name, country = map(str.lower, lc)
|
|
|
|
# if correct language-code is found
|
|
# set it as new search-language
|
|
if lang == lang_id\
|
|
or lang_id.startswith(lang)\
|
|
or lang == lang_name\
|
|
or lang.replace('_', ' ') == country:
|
|
parse_next = True
|
|
self.languages.append(lang)
|
|
break
|
|
|
|
# this force a engine or category
|
|
if query_part[0] == '!' or query_part[0] == '?':
|
|
prefix = query_part[1:].replace('_', ' ')
|
|
|
|
# check if prefix is equal with engine shortcut
|
|
if prefix in engine_shortcuts:
|
|
parse_next = True
|
|
self.engines.append({'category': 'none',
|
|
'name': engine_shortcuts[prefix]})
|
|
|
|
# check if prefix is equal with engine name
|
|
elif prefix in engines:
|
|
parse_next = True
|
|
self.engines.append({'category': 'none',
|
|
'name': prefix})
|
|
|
|
# check if prefix is equal with categorie name
|
|
elif prefix in categories:
|
|
# using all engines for that search, which
|
|
# are declared under that categorie name
|
|
parse_next = True
|
|
self.engines.extend({'category': prefix,
|
|
'name': engine.name}
|
|
for engine in categories[prefix]
|
|
if (engine.name, prefix) not in self.blocked_engines)
|
|
|
|
if query_part[0] == '!':
|
|
self.specific = True
|
|
|
|
# append query part to query_part list
|
|
self.query_parts.append(query_part)
|
|
|
|
def changeSearchQuery(self, search_query):
|
|
if len(self.query_parts):
|
|
self.query_parts[-1] = search_query
|
|
else:
|
|
self.query_parts.append(search_query)
|
|
|
|
def getSearchQuery(self):
|
|
if len(self.query_parts):
|
|
return self.query_parts[-1]
|
|
else:
|
|
return ''
|
|
|
|
def getFullQuery(self):
|
|
# get full querry including whitespaces
|
|
return string.join(self.query_parts, '')
|