2021-01-13 11:31:25 +01:00
|
|
|
# SPDX-License-Identifier: AGPL-3.0-or-later
|
|
|
|
|
2013-10-26 02:22:20 +02:00
|
|
|
from lxml import html
|
2020-10-02 18:13:56 +02:00
|
|
|
from urllib.parse import urlencode
|
2020-11-26 15:49:33 +01:00
|
|
|
from searx.utils import extract_text, extract_url, eval_xpath, eval_xpath_list
|
2013-10-26 02:22:20 +02:00
|
|
|
|
2014-01-20 02:31:20 +01:00
|
|
|
search_url = None
|
2022-06-06 00:18:33 +02:00
|
|
|
lang_all = 'en'
|
2014-01-20 02:31:20 +01:00
|
|
|
url_xpath = None
|
2013-10-26 02:22:20 +02:00
|
|
|
content_xpath = None
|
2014-01-20 02:31:20 +01:00
|
|
|
title_xpath = None
|
2019-07-25 07:46:41 +02:00
|
|
|
thumbnail_xpath = False
|
2016-11-30 18:43:03 +01:00
|
|
|
paging = False
|
2013-11-13 19:33:09 +01:00
|
|
|
suggestion_xpath = ''
|
2013-10-26 13:45:43 +02:00
|
|
|
results_xpath = ''
|
2016-05-19 07:38:43 +02:00
|
|
|
cached_xpath = ''
|
|
|
|
cached_url = ''
|
2021-05-17 15:04:55 +02:00
|
|
|
soft_max_redirects = 0
|
2013-10-26 02:22:20 +02:00
|
|
|
|
2022-06-06 00:18:33 +02:00
|
|
|
cookies = {}
|
|
|
|
headers = {}
|
|
|
|
'''Some engines might offer different result based on cookies or headers.
|
|
|
|
Possible use-case: To set safesearch cookie or header to moderate.'''
|
|
|
|
|
|
|
|
paging = False
|
|
|
|
'''Engine supports paging [True or False].'''
|
|
|
|
|
2016-03-28 15:15:03 +02:00
|
|
|
page_size = 1
|
|
|
|
# number of the first page (usually 0 or 1)
|
|
|
|
first_page_num = 1
|
|
|
|
|
2014-01-20 02:31:20 +01:00
|
|
|
|
2022-06-06 00:18:33 +02:00
|
|
|
time_range_support = False
|
|
|
|
'''Engine supports search time range.'''
|
|
|
|
|
|
|
|
time_range_url = '&hours={time_range_val}'
|
|
|
|
'''Time range URL parameter in the in :py:obj:`search_url`. If no time range is
|
2022-09-29 23:06:59 +02:00
|
|
|
requested by the user, the URL parameter is an empty string. The
|
2022-06-06 00:18:33 +02:00
|
|
|
``{time_range_val}`` replacement is taken from the :py:obj:`time_range_map`.
|
|
|
|
|
|
|
|
.. code:: yaml
|
|
|
|
|
|
|
|
time_range_url : '&days={time_range_val}'
|
|
|
|
'''
|
|
|
|
|
|
|
|
time_range_map = {
|
|
|
|
'day': 24,
|
|
|
|
'week': 24 * 7,
|
|
|
|
'month': 24 * 30,
|
|
|
|
'year': 24 * 365,
|
|
|
|
}
|
|
|
|
'''Maps time range value from user to ``{time_range_val}`` in
|
|
|
|
:py:obj:`time_range_url`.
|
|
|
|
|
|
|
|
.. code:: yaml
|
|
|
|
|
|
|
|
time_range_map:
|
|
|
|
day: 1
|
|
|
|
week: 7
|
|
|
|
month: 30
|
|
|
|
year: 365
|
|
|
|
'''
|
|
|
|
|
|
|
|
safe_search_support = False
|
|
|
|
'''Engine supports safe-search.'''
|
|
|
|
|
|
|
|
safe_search_map = {
|
|
|
|
0: '&filter=none',
|
|
|
|
1: '&filter=moderate',
|
|
|
|
2: '&filter=strict'
|
|
|
|
}
|
|
|
|
'''Maps safe-search value to ``{safe_search}`` in :py:obj:`search_url`.
|
|
|
|
|
|
|
|
.. code:: yaml
|
|
|
|
|
|
|
|
safesearch: true
|
|
|
|
safes_search_map:
|
|
|
|
0: '&filter=none'
|
|
|
|
1: '&filter=moderate'
|
|
|
|
2: '&filter=strict'
|
|
|
|
|
|
|
|
'''
|
|
|
|
|
|
|
|
|
2013-10-26 02:22:20 +02:00
|
|
|
def request(query, params):
|
|
|
|
query = urlencode({'q': query})[2:]
|
2016-03-28 15:15:03 +02:00
|
|
|
|
|
|
|
fp = {'query': query}
|
|
|
|
if paging and search_url.find('{pageno}') >= 0:
|
2016-08-14 13:46:54 +02:00
|
|
|
fp['pageno'] = (params['pageno'] - 1) * page_size + first_page_num
|
2016-03-28 15:15:03 +02:00
|
|
|
|
2022-06-06 00:18:33 +02:00
|
|
|
safe_search = ''
|
|
|
|
if params['safesearch']:
|
|
|
|
safe_search = safe_search_map[params['safesearch']]
|
|
|
|
|
|
|
|
lang = lang_all
|
|
|
|
if params['language'] != 'all':
|
|
|
|
lang = params['language'][:2]
|
|
|
|
|
|
|
|
time_range = ''
|
|
|
|
if params.get('time_range'):
|
|
|
|
time_range_val = time_range_map.get(params.get('time_range'))
|
|
|
|
time_range = time_range_url.format(time_range_val=time_range_val)
|
|
|
|
|
|
|
|
safe_search = ''
|
|
|
|
if params['safesearch']:
|
|
|
|
safe_search = safe_search_map[params['safesearch']]
|
|
|
|
|
|
|
|
fargs = {
|
|
|
|
'query': urlencode({'q': query})[2:],
|
|
|
|
'lang': lang,
|
|
|
|
'pageno': (params['pageno'] - 1) * page_size + first_page_num,
|
|
|
|
'time_range': time_range,
|
|
|
|
'safe_search': safe_search,
|
|
|
|
}
|
|
|
|
|
|
|
|
params['cookies'].update(cookies)
|
|
|
|
params['headers'].update(headers)
|
|
|
|
|
|
|
|
params['url'] = search_url.format(**fargs)
|
2021-05-17 15:04:55 +02:00
|
|
|
params['soft_max_redirects'] = soft_max_redirects
|
2016-03-28 15:15:03 +02:00
|
|
|
|
2013-10-26 02:22:20 +02:00
|
|
|
return params
|
|
|
|
|
|
|
|
|
|
|
|
def response(resp):
|
|
|
|
results = []
|
|
|
|
dom = html.fromstring(resp.text)
|
2020-12-10 10:40:45 +01:00
|
|
|
is_onion = True if 'onions' in categories else False # pylint: disable=undefined-variable
|
2016-05-19 07:38:43 +02:00
|
|
|
|
2013-10-26 13:45:43 +02:00
|
|
|
if results_xpath:
|
2020-11-26 15:49:33 +01:00
|
|
|
for result in eval_xpath_list(dom, results_xpath):
|
|
|
|
url = extract_url(eval_xpath_list(result, url_xpath, min_len=1), search_url)
|
|
|
|
title = extract_text(eval_xpath_list(result, title_xpath, min_len=1))
|
2021-09-04 12:41:23 +02:00
|
|
|
content = extract_text(eval_xpath_list(result, content_xpath))
|
2019-07-25 07:46:41 +02:00
|
|
|
tmp_result = {'url': url, 'title': title, 'content': content}
|
|
|
|
|
|
|
|
# add thumbnail if available
|
|
|
|
if thumbnail_xpath:
|
2020-11-26 15:49:33 +01:00
|
|
|
thumbnail_xpath_result = eval_xpath_list(result, thumbnail_xpath)
|
2019-07-25 09:31:47 +02:00
|
|
|
if len(thumbnail_xpath_result) > 0:
|
|
|
|
tmp_result['img_src'] = extract_url(thumbnail_xpath_result, search_url)
|
2019-07-25 07:46:41 +02:00
|
|
|
|
2016-05-19 07:38:43 +02:00
|
|
|
# add alternative cached url if available
|
|
|
|
if cached_xpath:
|
2020-11-26 15:49:33 +01:00
|
|
|
tmp_result['cached_url'] = cached_url\
|
|
|
|
+ extract_text(eval_xpath_list(result, cached_xpath, min_len=1))
|
2016-05-19 07:38:43 +02:00
|
|
|
|
|
|
|
if is_onion:
|
|
|
|
tmp_result['is_onion'] = True
|
|
|
|
|
2019-07-25 07:46:41 +02:00
|
|
|
results.append(tmp_result)
|
2013-10-26 13:45:43 +02:00
|
|
|
else:
|
2016-05-19 07:38:43 +02:00
|
|
|
if cached_xpath:
|
|
|
|
for url, title, content, cached in zip(
|
|
|
|
(extract_url(x, search_url) for
|
2020-11-26 15:49:33 +01:00
|
|
|
x in eval_xpath_list(dom, url_xpath)),
|
|
|
|
map(extract_text, eval_xpath_list(dom, title_xpath)),
|
|
|
|
map(extract_text, eval_xpath_list(dom, content_xpath)),
|
|
|
|
map(extract_text, eval_xpath_list(dom, cached_xpath))
|
2016-05-19 07:38:43 +02:00
|
|
|
):
|
|
|
|
results.append({'url': url, 'title': title, 'content': content,
|
|
|
|
'cached_url': cached_url + cached, 'is_onion': is_onion})
|
|
|
|
else:
|
|
|
|
for url, title, content in zip(
|
|
|
|
(extract_url(x, search_url) for
|
2020-11-26 15:49:33 +01:00
|
|
|
x in eval_xpath_list(dom, url_xpath)),
|
|
|
|
map(extract_text, eval_xpath_list(dom, title_xpath)),
|
|
|
|
map(extract_text, eval_xpath_list(dom, content_xpath))
|
2016-05-19 07:38:43 +02:00
|
|
|
):
|
|
|
|
results.append({'url': url, 'title': title, 'content': content, 'is_onion': is_onion})
|
2013-10-26 13:45:43 +02:00
|
|
|
|
2013-11-13 19:33:09 +01:00
|
|
|
if not suggestion_xpath:
|
|
|
|
return results
|
2019-11-15 09:31:37 +01:00
|
|
|
for suggestion in eval_xpath(dom, suggestion_xpath):
|
2014-01-05 14:06:52 +01:00
|
|
|
results.append({'suggestion': extract_text(suggestion)})
|
2013-10-26 02:22:20 +02:00
|
|
|
return results
|