on master help sign in
Merge pull request #2398 from dalf/mod-search-query

Mod search query

commit04447f8c1a32aa2d1a2209d0a120ce04435c1346 authorAlexandre Flament <alex@al-f.net> dateSun 20 Dec 2020, 09:32 (6 years ago) parent35feb00eda8934
5 files changed+36−53
searx/query.py+2−2 searx/search/__init__.py+15−15 searx/webadapter.py+6−23 tests/unit/test_search.py+11−11 tests/unit/test_standalone_searx.py+2−2
searx/query.py+2−2
@@ -125,12 +125,12 @@ class RawTextQuery:
125125 searx_query_part = True
126126 engine_name = engine_shortcuts[prefix]
127127 if engine_name in engines:
128− self.enginerefs.append(EngineRef(engine_name, 'none', True))
128+ self.enginerefs.append(EngineRef(engine_name, 'none'))
129129
130130 # check if prefix is equal with engine name
131131 elif prefix in engines:
132132 searx_query_part = True
133− self.enginerefs.append(EngineRef(prefix, 'none', True))
133+ self.enginerefs.append(EngineRef(prefix, 'none'))
134134
135135 # check if prefix is equal with categorie name
136136 elif prefix in categories:
searx/search/__init__.py+15−15
@@ -52,39 +52,36 @@ def initialize(settings_engines=None):
5252
5353class EngineRef:
5454
55− __slots__ = 'name', 'category', 'from_bang'
55+ __slots__ = 'name', 'category'
5656
57− def __init__(self, name: str, category: str, from_bang: bool=False):
57+ def __init__(self, name: str, category: str):
5858 self.name = name
5959 self.category = category
60− self.from_bang = from_bang
6160
6261 def __repr__(self):
63− return "EngineRef({!r}, {!r}, {!r})".format(self.name, self.category, self.from_bang)
62+ return "EngineRef({!r}, {!r})".format(self.name, self.category)
6463
6564 def __eq__(self, other):
66− return self.name == other.name and self.category == other.category and self.from_bang == other.from_bang
65+ return self.name == other.name and self.category == other.category
6766
6867
6968class SearchQuery:
7069 """container for all the search parameters (query, language, etc...)"""
7170
72− __slots__ = 'query', 'engineref_list', 'categories', 'lang', 'safesearch', 'pageno', 'time_range',\
71+ __slots__ = 'query', 'engineref_list', 'lang', 'safesearch', 'pageno', 'time_range',\
7372 'timeout_limit', 'external_bang'
7473
7574 def __init__(self,
7675 query: str,
7776 engineref_list: typing.List[EngineRef],
78− categories: typing.List[str],
79− lang: str,
80− safesearch: int,
81− pageno: int,
82− time_range: typing.Optional[str],
77+ lang: str='all',
78+ safesearch: int=0,
79+ pageno: int=1,
80+ time_range: typing.Optional[str]=None,
8381 timeout_limit: typing.Optional[float]=None,
8482 external_bang: typing.Optional[str]=None):
8583 self.query = query
8684 self.engineref_list = engineref_list
87− self.categories = categories
8885 self.lang = lang
8986 self.safesearch = safesearch
9087 self.pageno = pageno
@@ -92,15 +89,18 @@ class SearchQuery:
9289 self.timeout_limit = timeout_limit
9390 self.external_bang = external_bang
9491
92+ @property
93+ def categories(self):
94+ return list(set(map(lambda engineref: engineref.category, self.engineref_list)))
95+
9596 def __repr__(self):
96− return "SearchQuery({!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r})".\
97− format(self.query, self.engineref_list, self.categories, self.lang, self.safesearch,
97+ return "SearchQuery({!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r}, {!r})".\
98+ format(self.query, self.engineref_list, self.lang, self.safesearch,
9899 self.pageno, self.time_range, self.timeout_limit, self.external_bang)
99100
100101 def __eq__(self, other):
101102 return self.query == other.query\
102103 and self.engineref_list == other.engineref_list\
103− and self.categories == self.categories\
104104 and self.lang == other.lang\
105105 and self.safesearch == other.safesearch\
106106 and self.pageno == other.pageno\
searx/webadapter.py+6−23
@@ -109,18 +109,6 @@ def parse_timeout(form: Dict[str, str], raw_text_query: RawTextQuery) -> Optiona
109109 raise SearxParameterException('timeout_limit', timeout_limit)
110110
111111
112−def parse_specific(raw_text_query: RawTextQuery) -> Tuple[List[EngineRef], List[str]]:
113− query_engineref_list = raw_text_query.enginerefs
114− additional_categories = set()
115− for engineref in raw_text_query.enginerefs:
116− if engineref.from_bang:
117− additional_categories.add('none')
118− else:
119− additional_categories.add(engineref.category)
120− query_categories = list(additional_categories)
121− return query_engineref_list, query_categories
122−
123−
124112def parse_category_form(query_categories: List[str], name: str, value: str) -> None:
125113 if name == 'categories':
126114 query_categories.extend(categ for categ in map(str.strip, value.split(',')) if categ in categories)
@@ -171,8 +159,7 @@ def get_engineref_from_category_list(category_list: List[str], disabled_engines:
171159 return result
172160
173161
174−def parse_generic(preferences: Preferences, form: Dict[str, str], disabled_engines: List[str])\
175− -> Tuple[List[EngineRef], List[str]]:
162+def parse_generic(preferences: Preferences, form: Dict[str, str], disabled_engines: List[str]) -> List[EngineRef]:
176163 query_engineref_list = []
177164 query_categories = []
178165
@@ -195,8 +182,6 @@ def parse_generic(preferences: Preferences, form: Dict[str, str], disabled_engin
195182 if query_categories:
196183 # add engines from referenced by the "categories" parameter and the "category_*"" parameters
197184 query_engineref_list.extend(get_engineref_from_category_list(query_categories, disabled_engines))
198− # get categories from the query_engineref_list
199− query_categories = list(set(engine.category for engine in query_engineref_list))
200185 else:
201186 # no "engines" parameters in the form
202187 if not query_categories:
@@ -208,7 +193,7 @@ def parse_generic(preferences: Preferences, form: Dict[str, str], disabled_engin
208193 # declared under the specific categories
209194 query_engineref_list.extend(get_engineref_from_category_list(query_categories, disabled_engines))
210195
211− return query_engineref_list, query_categories
196+ return query_engineref_list
212197
213198
214199def get_search_query_from_webapp(preferences: Preferences, form: Dict[str, str])\
@@ -236,20 +221,18 @@ def get_search_query_from_webapp(preferences: Preferences, form: Dict[str, str])
236221 if not is_locked('categories') and raw_text_query.enginerefs and raw_text_query.specific:
237222 # if engines are calculated from query,
238223 # set categories by using that informations
239− query_engineref_list, query_categories = parse_specific(raw_text_query)
224+ query_engineref_list = raw_text_query.enginerefs
240225 else:
241226 # otherwise, using defined categories to
242227 # calculate which engines should be used
243− query_engineref_list, query_categories = parse_generic(preferences, form, disabled_engines)
228+ query_engineref_list = parse_generic(preferences, form, disabled_engines)
244229
245230 query_engineref_list = deduplicate_engineref_list(query_engineref_list)
246231 query_engineref_list, query_engineref_list_unknown, query_engineref_list_notoken =\
247232 validate_engineref_list(query_engineref_list, preferences)
248233
249− return (SearchQuery(query, query_engineref_list, query_categories,
250− query_lang, query_safesearch, query_pageno,
251− query_time_range, query_timeout,
252− external_bang=external_bang),
234+ return (SearchQuery(query, query_engineref_list, query_lang, query_safesearch, query_pageno,
235+ query_time_range, query_timeout, external_bang=external_bang),
253236 raw_text_query,
254237 query_engineref_list_unknown,
255238 query_engineref_list_notoken)
tests/unit/test_search.py+11−11
@@ -23,13 +23,13 @@ TEST_ENGINES = [
2323class SearchQueryTestCase(SearxTestCase):
2424
2525 def test_repr(self):
26− s = SearchQuery('test', [EngineRef('bing', 'general', False)], ['general'], 'all', 0, 1, '1', 5.0, 'g')
26+ s = SearchQuery('test', [EngineRef('bing', 'general')], 'all', 0, 1, '1', 5.0, 'g')
2727 self.assertEqual(repr(s),
28− "SearchQuery('test', [EngineRef('bing', 'general', False)], ['general'], 'all', 0, 1, '1', 5.0, 'g')") # noqa
28+ "SearchQuery('test', [EngineRef('bing', 'general')], 'all', 0, 1, '1', 5.0, 'g')") # noqa
2929
3030 def test_eq(self):
31− s = SearchQuery('test', [EngineRef('bing', 'general', False)], ['general'], 'all', 0, 1, None, None, None)
32− t = SearchQuery('test', [EngineRef('google', 'general', False)], ['general'], 'all', 0, 1, None, None, None)
31+ s = SearchQuery('test', [EngineRef('bing', 'general')], 'all', 0, 1, None, None, None)
32+ t = SearchQuery('test', [EngineRef('google', 'general')], 'all', 0, 1, None, None, None)
3333 self.assertEqual(s, s)
3434 self.assertNotEqual(s, t)
3535
@@ -43,7 +43,7 @@ class SearchTestCase(SearxTestCase):
4343 def test_timeout_simple(self):
4444 searx.search.max_request_timeout = None
4545 search_query = SearchQuery('test', [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
46− ['general'], 'en-US', SAFESEARCH, PAGENO, None, None)
46+ 'en-US', SAFESEARCH, PAGENO, None, None)
4747 search = searx.search.Search(search_query)
4848 search.search()
4949 self.assertEqual(search.actual_timeout, 3.0)
@@ -51,7 +51,7 @@ class SearchTestCase(SearxTestCase):
5151 def test_timeout_query_above_default_nomax(self):
5252 searx.search.max_request_timeout = None
5353 search_query = SearchQuery('test', [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
54− ['general'], 'en-US', SAFESEARCH, PAGENO, None, 5.0)
54+ 'en-US', SAFESEARCH, PAGENO, None, 5.0)
5555 search = searx.search.Search(search_query)
5656 search.search()
5757 self.assertEqual(search.actual_timeout, 3.0)
@@ -59,7 +59,7 @@ class SearchTestCase(SearxTestCase):
5959 def test_timeout_query_below_default_nomax(self):
6060 searx.search.max_request_timeout = None
6161 search_query = SearchQuery('test', [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
62− ['general'], 'en-US', SAFESEARCH, PAGENO, None, 1.0)
62+ 'en-US', SAFESEARCH, PAGENO, None, 1.0)
6363 search = searx.search.Search(search_query)
6464 search.search()
6565 self.assertEqual(search.actual_timeout, 1.0)
@@ -67,7 +67,7 @@ class SearchTestCase(SearxTestCase):
6767 def test_timeout_query_below_max(self):
6868 searx.search.max_request_timeout = 10.0
6969 search_query = SearchQuery('test', [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
70− ['general'], 'en-US', SAFESEARCH, PAGENO, None, 5.0)
70+ 'en-US', SAFESEARCH, PAGENO, None, 5.0)
7171 search = searx.search.Search(search_query)
7272 search.search()
7373 self.assertEqual(search.actual_timeout, 5.0)
@@ -75,7 +75,7 @@ class SearchTestCase(SearxTestCase):
7575 def test_timeout_query_above_max(self):
7676 searx.search.max_request_timeout = 10.0
7777 search_query = SearchQuery('test', [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
78− ['general'], 'en-US', SAFESEARCH, PAGENO, None, 15.0)
78+ 'en-US', SAFESEARCH, PAGENO, None, 15.0)
7979 search = searx.search.Search(search_query)
8080 search.search()
8181 self.assertEqual(search.actual_timeout, 10.0)
@@ -83,7 +83,7 @@ class SearchTestCase(SearxTestCase):
8383 def test_external_bang(self):
8484 search_query = SearchQuery('yes yes',
8585 [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
86− ['general'], 'en-US', SAFESEARCH, PAGENO, None, None,
86+ 'en-US', SAFESEARCH, PAGENO, None, None,
8787 external_bang="yt")
8888 search = searx.search.Search(search_query)
8989 results = search.search()
@@ -92,7 +92,7 @@ class SearchTestCase(SearxTestCase):
9292
9393 search_query = SearchQuery('youtube never gonna give you up',
9494 [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
95− ['general'], 'en-US', SAFESEARCH, PAGENO, None, None)
95+ 'en-US', SAFESEARCH, PAGENO, None, None)
9696
9797 search = searx.search.Search(search_query)
9898 results = search.search()
tests/unit/test_standalone_searx.py+2−2
@@ -101,8 +101,8 @@ class StandaloneSearx(SearxTestCase):
101101 args = sas.parse_argument(['rain', ])
102102 search_q = sas.get_search_query(args)
103103 self.assertTrue(search_q)
104− self.assertEqual(search_q, SearchQuery('rain', [EngineRef('engine1', 'general', False)],
105− ['general'], 'all', 0, 1, None, None, None))
104+ self.assertEqual(search_q, SearchQuery('rain', [EngineRef('engine1', 'general')],
105+ 'all', 0, 1, None, None, None))
106106
107107 def test_no_parsed_url(self):
108108 """test no_parsed_url func"""