zaclys/searxng
1
0
Fork 1
mirror of https://github.com/searxng/searxng synced 2024-01-01 19:24:07 +01:00
Code Issues Projects Releases Packages Wiki Activity

[mod] remove searxng_extra/standalone_searx.py

Browse source 
Script did not work / is not maintained.

Closes: https://github.com/searxng/searxng/issues/332
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
This commit is contained in:
Markus Heiser 2021-12-24 07:35:30 +01:00
parent 69747d5cb3
commit f77fc8132b
5 changed files with 0 additions and 341 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
docs/index.rst
View file

@ -33,7 +33,6 @@ If you don't trust anyone, you can set up your own, see :ref:`installation`.
user/index
admin/index
dev/index
searxng_extra/index
utils/index
src/index

14
docs/searxng_extra/index.rst
View file

@ -1,14 +0,0 @@
.. _searxng_extra:
======================================================
Tooling box ``searxng_extra`` for developers and users
======================================================
In the folder :origin:`searxng_extra/` we maintain some tools useful for
developers and users.
.. toctree::
:maxdepth: 2
:caption: Contents
standalone_searx.py

9
docs/searxng_extra/standalone_searx.py.rst
View file

@ -1,9 +0,0 @@
.. _standalone_searx.py:
=====================================
``searxng_extra/standalone_searx.py``
=====================================
.. automodule:: searxng_extra.standalone_searx
:members:

206
searxng_extra/standalone_searx.py
View file

@ -1,206 +0,0 @@
#!/usr/bin/env python
# lint: pylint
# SPDX-License-Identifier: AGPL-3.0-or-later
# (C) Copyright Contributors to the SearXNG project.
# (C) Copyright Contributors to the searx project (2014 - 2021)
"""Script to run SearXNG from terminal.
Getting categories without initiate the engine will only return `['general']`
>>> import searx.engines
... list(searx.engines.categories.keys())
['general']
>>> import searx.search
... searx.search.initialize()
... list(searx.engines.categories.keys())
['general', 'it', 'science', 'images', 'news', 'videos', 'music', 'files', 'social media', 'map']
Example to use this script:
.. code:: bash
$ python3 searxng_extra/standalone_searx.py rain
Example to run it from python:
>>> import importlib
... import json
... import sys
... import searx.engines
... import searx.search
... search_query = 'rain'
... # initialize engines
... searx.search.initialize()
... # load engines categories once instead of each time the function called
... engine_cs = list(searx.engines.categories.keys())
... # load module
... spec = importlib.util.spec_from_file_location(
... 'utils.standalone_searx', 'searxng_extra/standalone_searx.py')
... sas = importlib.util.module_from_spec(spec)
... spec.loader.exec_module(sas)
... # use function from module
... prog_args = sas.parse_argument([search_query], category_choices=engine_cs)
... search_q = sas.get_search_query(prog_args, engine_categories=engine_cs)
... res_dict = sas.to_dict(search_q)
... sys.stdout.write(json.dumps(
... res_dict, sort_keys=True, indent=4, ensure_ascii=False,
... default=sas.json_serial))
{
"answers": [],
"infoboxes": [ {...} ],
"paging": true,
"results": [... ],
"results_number": 820000000.0,
"search": {
"lang": "all",
"pageno": 1,
"q": "rain",
"safesearch": 0,
"timerange": null
},
"suggestions": [...]
}
""" # pylint: disable=line-too-long
import argparse
import sys
from datetime import datetime
from json import dumps
from typing import Any, Dict, List, Optional
import searx
import searx.preferences
import searx.query
import searx.search
import searx.webadapter
EngineCategoriesVar = Optional[List[str]]
def get_search_query(
args: argparse.Namespace, engine_categories: EngineCategoriesVar = None
) -> searx.search.SearchQuery:
"""Get search results for the query"""
if engine_categories is None:
engine_categories = list(searx.engines.categories.keys())
try:
category = args.category.decode('utf-8')
except AttributeError:
category = args.category
form = {
"q": args.query,
"categories": category,
"pageno": str(args.pageno),
"language": args.lang,
"time_range": args.timerange
}
preferences = searx.preferences.Preferences(
['oscar'], engine_categories, searx.engines.engines, [])
preferences.key_value_settings['safesearch'].parse(args.safesearch)
search_query = searx.webadapter.get_search_query_from_webapp(
preferences, form)[0]
return search_query
def no_parsed_url(results: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
"""Remove parsed url from dict."""
for result in results:
del result['parsed_url']
return results
def json_serial(obj: Any) -> Any:
"""JSON serializer for objects not serializable by default json code.
:raise TypeError: raised when **obj** is not serializable
"""
if isinstance(obj, datetime):
serial = obj.isoformat()
return serial
if isinstance(obj, bytes):
return obj.decode('utf8')
if isinstance(obj, set):
return list(obj)
raise TypeError("Type ({}) not serializable".format(type(obj)))
def to_dict(search_query: searx.search.SearchQuery) -> Dict[str, Any]:
"""Get result from parsed arguments."""
result_container = searx.search.Search(search_query).search()
result_container_json = {
"search": {
"q": search_query.query,
"pageno": search_query.pageno,
"lang": search_query.lang,
"safesearch": search_query.safesearch,
"timerange": search_query.time_range,
},
"results": no_parsed_url(result_container.get_ordered_results()),
"infoboxes": result_container.infoboxes,
"suggestions": list(result_container.suggestions),
"answers": list(result_container.answers),
"paging": result_container.paging,
"results_number": result_container.results_number()
}
return result_container_json
def parse_argument(
args: Optional[List[str]]=None,
category_choices: EngineCategoriesVar=None
) -> argparse.Namespace:
"""Parse command line.
:raise SystemExit: Query argument required on `args`
Examples:
>>> import importlib
... # load module
... spec = importlib.util.spec_from_file_location(
... 'utils.standalone_searx', 'utils/standalone_searx.py')
... sas = importlib.util.module_from_spec(spec)
... spec.loader.exec_module(sas)
... sas.parse_argument()
usage: ptipython [-h] [--category [{general}]] [--lang [LANG]] [--pageno [PAGENO]] [--safesearch [{0,1,2}]] [--timerange [{day,week,month,year}]]
query
SystemExit: 2
>>> sas.parse_argument(['rain'])
Namespace(category='general', lang='all', pageno=1, query='rain', safesearch='0', timerange=None)
""" # noqa: E501
if not category_choices:
category_choices = list(searx.engines.categories.keys())
parser = argparse.ArgumentParser(description='Standalone searx.')
parser.add_argument('query', type=str,
help='Text query')
parser.add_argument('--category', type=str, nargs='?',
choices=category_choices,
default='general',
help='Search category')
parser.add_argument('--lang', type=str, nargs='?', default='all',
help='Search language')
parser.add_argument('--pageno', type=int, nargs='?', default=1,
help='Page number starting from 1')
parser.add_argument(
'--safesearch', type=str, nargs='?',
choices=['0', '1', '2'], default='0',
help='Safe content filter from none to strict')
parser.add_argument(
'--timerange', type=str,
nargs='?', choices=['day', 'week', 'month', 'year'],
help='Filter by time range')
return parser.parse_args(args)
if __name__ == '__main__':
searx.search.initialize()
engine_cs = list(searx.engines.categories.keys())
prog_args = parse_argument(category_choices=engine_cs)
search_q = get_search_query(prog_args, engine_categories=engine_cs)
res_dict = to_dict(search_q)
sys.stdout.write(dumps(
res_dict, sort_keys=True, indent=4, ensure_ascii=False,
default=json_serial))

111
tests/unit/test_standalone_searx.py
View file

@ -1,111 +0,0 @@
# -*- coding: utf-8 -*-
"""Test utils/standalone_searx.py"""
import datetime
import io
import sys
from mock import Mock, patch
from nose2.tools import params
from searx.search import SearchQuery, EngineRef, initialize
from searxng_extra import standalone_searx as sas
from tests import SearxTestCase
class StandaloneSearx(SearxTestCase):
"""Unit test for standalone_searx."""
@classmethod
def setUpClass(cls):
engine_list = [{'engine': 'dummy', 'name': 'engine1', 'shortcut': 'e1'}]
initialize(engine_list)
def test_parse_argument_no_args(self):
"""Test parse argument without args."""
with patch.object(sys, 'argv', ['standalone_searx']), \
self.assertRaises(SystemExit):
sys.stderr = io.StringIO()
sas.parse_argument()
sys.stdout = sys.__stderr__
def test_parse_argument_basic_args(self):
"""Test parse argument with basic args."""
query = 'red box'
exp_dict = {
'query': query, 'category': 'general', 'lang': 'all', 'pageno': 1,
'safesearch': '0', 'timerange': None}
args = ['standalone_searx', query]
with patch.object(sys, 'argv', args):
res = sas.parse_argument()
self.assertEqual(exp_dict, vars(res))
res2 = sas.parse_argument(args[1:])
self.assertEqual(exp_dict, vars(res2))
def test_to_dict(self):
"""test to_dict."""
self.assertEqual(
sas.to_dict(
sas.get_search_query(sas.parse_argument(['red box']))),
{
'search': {
'q': 'red box', 'pageno': 1, 'lang': 'all',
'safesearch': 0, 'timerange': None
},
'results': [], 'infoboxes': [], 'suggestions': [],
'answers': [], 'paging': False, 'results_number': 0
}
)
def test_to_dict_with_mock(self):
"""test to dict."""
with patch.object(sas.searx.search, 'Search') as mock_s:
m_search = mock_s().search()
m_sq = Mock()
self.assertEqual(
sas.to_dict(m_sq),
{
'answers': [],
'infoboxes': m_search.infoboxes,
'paging': m_search.paging,
'results': m_search.get_ordered_results(),
'results_number': m_search.results_number(),
'search': {
'lang': m_sq.lang,
'pageno': m_sq.pageno,
'q': m_sq.query,
'safesearch': m_sq.safesearch,
'timerange': m_sq.time_range,
},
'suggestions': []
}
)
def test_get_search_query(self):
"""test get_search_query."""
args = sas.parse_argument(['rain', ])
search_q = sas.get_search_query(args)
self.assertTrue(search_q)
self.assertEqual(search_q, SearchQuery('rain', [EngineRef('engine1', 'general')],
'all', 0, 1, None, None, None))
def test_no_parsed_url(self):
"""test no_parsed_url func"""
self.assertEqual(
sas.no_parsed_url([{'parsed_url': 'http://example.com'}]),
[{}]
)
@params(
(datetime.datetime(2020, 1, 1), '2020-01-01T00:00:00'),
('a'.encode('utf8'), 'a'),
(set([1]), [1])
)
def test_json_serial(self, arg, exp_res):
"""test json_serial func"""
self.assertEqual(sas.json_serial(arg), exp_res)
def test_json_serial_error(self):
"""test error on json_serial."""
with self.assertRaises(TypeError):
sas.json_serial('a')