on master help sign in
[mod] add searx.search.EngineRef
2929495Alexandre Flament
searx/query.py+8−14
@@ -20,9 +20,8 @@ along with searx. If not, see < http://www.gnu.org/licenses/ >.
20import re
20import re
21
21
22from searx.languages import language_codes
22from searx.languages import language_codes
23−from searx.engines import (
24− categories, engines, engine_shortcuts
23+from searx.engines import categories, engines, engine_shortcuts
25−)
24+from searx.search import EngineRef
26
25
27
26
28VALID_LANGUAGE_CODE = re.compile(r'^[a-z]{2,3}(-[a-zA-Z]{2})?$')
27VALID_LANGUAGE_CODE = re.compile(r'^[a-z]{2,3}(-[a-zA-Z]{2})?$')
@@ -40,7 +39,7 @@ class RawTextQuery:
40 self.disabled_engines = disabled_engines
39 self.disabled_engines = disabled_engines
41
40
42 self.query_parts = []
41 self.query_parts = []
43− self.engines = []
42+ self.enginerefs = []
44 self.languages = []
43 self.languages = []
45 self.timeout_limit = None
44 self.timeout_limit = None
46 self.external_bang = None
45 self.external_bang = None
@@ -135,26 +134,21 @@ class RawTextQuery:
135 parse_next = True
134 parse_next = True
136 engine_name = engine_shortcuts[prefix]
135 engine_name = engine_shortcuts[prefix]
137 if engine_name in engines:
136 if engine_name in engines:
138− self.engines.append({'category': 'none',
137+ self.enginerefs.append(EngineRef(engine_name, 'none', True))
139− 'name': engine_name,
140− 'from_bang': True})
141
138
142 # check if prefix is equal with engine name
139 # check if prefix is equal with engine name
143 elif prefix in engines:
140 elif prefix in engines:
144 parse_next = True
141 parse_next = True
145− self.engines.append({'category': 'none',
142+ self.enginerefs.append(EngineRef(prefix, 'none', True))
146− 'name': prefix,
147− 'from_bang': True})
148
143
149 # check if prefix is equal with categorie name
144 # check if prefix is equal with categorie name
150 elif prefix in categories:
145 elif prefix in categories:
151 # using all engines for that search, which
146 # using all engines for that search, which
152 # are declared under that categorie name
147 # are declared under that categorie name
153 parse_next = True
148 parse_next = True
154− self.engines.extend({'category': prefix,
149+ self.enginerefs.extend(EngineRef(engine.name, prefix)
155− 'name': engine.name}
156− for engine in categories[prefix]
150+ for engine in categories[prefix]
157− if (engine.name, prefix) not in self.disabled_engines)
151+ if (engine.name, prefix) not in self.disabled_engines)
158
152
159 if query_part[0] == '!':
153 if query_part[0] == '!':
160 self.specific = True
154 self.specific = True
searx/search.py+22−13
@@ -47,13 +47,24 @@ else:
47 exit(1)
47 exit(1)
48
48
49
49
50+class EngineRef:
51+
52+ def __init__(self, name, category, from_bang=False):
53+ self.name = name
54+ self.category = category
55+ self.from_bang = from_bang
56+
57+ def __str__(self):
58+ return "(" + self.name + "," + self.category + "," + str(self.from_bang) + ")"
59+
60+
50class SearchQuery:
61class SearchQuery:
51 """container for all the search parameters (query, language, etc...)"""
62 """container for all the search parameters (query, language, etc...)"""
52
63
53− def __init__(self, query, engines, categories, lang, safesearch, pageno, time_range,
64+ def __init__(self, query, engineref_list, categories, lang, safesearch, pageno, time_range,
54 timeout_limit=None, preferences=None, external_bang=None):
65 timeout_limit=None, preferences=None, external_bang=None):
55 self.query = query
66 self.query = query
56− self.engines = engines
67+ self.engineref_list = engineref_list
57 self.categories = categories
68 self.categories = categories
58 self.lang = lang
69 self.lang = lang
59 self.safesearch = safesearch
70 self.safesearch = safesearch
@@ -64,7 +75,7 @@ class SearchQuery:
64 self.external_bang = external_bang
75 self.external_bang = external_bang
65
76
66 def __str__(self):
77 def __str__(self):
67− return self.query + ";" + str(self.engines)
78+ return self.query + ";" + str(self.engineref_list)
68
79
69
80
70def send_http_request(engine, request_params):
81def send_http_request(engine, request_params):
@@ -318,13 +329,13 @@ class Search:
318
329
319 return True
330 return True
320
331
321− def _get_params(self, selected_engine, user_agent):
332+ def _get_params(self, engineref, user_agent):
322− if selected_engine['name'] not in engines:
333+ if engineref.name not in engines:
323 return None, None
334 return None, None
324
335
325− engine = engines[selected_engine['name']]
336+ engine = engines[engineref.name]
326
337
327− if not self._is_accepted(selected_engine['name'], engine):
338+ if not self._is_accepted(engineref.name, engine):
328 return None, None
339 return None, None
329
340
330 # set default request parameters
341 # set default request parameters
@@ -341,15 +352,13 @@ class Search:
341 request_params['safesearch'] = self.search_query.safesearch
352 request_params['safesearch'] = self.search_query.safesearch
342 request_params['time_range'] = self.search_query.time_range
353 request_params['time_range'] = self.search_query.time_range
343
354
344− request_params['category'] = selected_engine['category']
355+ request_params['category'] = engineref.category
345 request_params['pageno'] = self.search_query.pageno
356 request_params['pageno'] = self.search_query.pageno
346
357
347 return request_params, engine.timeout
358 return request_params, engine.timeout
348
359
349 # do search-request
360 # do search-request
350 def _get_requests(self):
361 def _get_requests(self):
351− global number_of_searches
352−
353 # init vars
362 # init vars
354 requests = []
363 requests = []
355
364
@@ -361,14 +370,14 @@ class Search:
361 default_timeout = 0
370 default_timeout = 0
362
371
363 # start search-reqest for all selected engines
372 # start search-reqest for all selected engines
364− for selected_engine in self.search_query.engines:
373+ for engineref in self.search_query.engineref_list:
365 # set default request parameters
374 # set default request parameters
366− request_params, engine_timeout = self._get_params(selected_engine, user_agent)
375+ request_params, engine_timeout = self._get_params(engineref, user_agent)
367 if request_params is None:
376 if request_params is None:
368 continue
377 continue
369
378
370 # append request to list
379 # append request to list
371− requests.append((selected_engine['name'], self.search_query.query, request_params))
380+ requests.append((engineref.name, self.search_query.query, request_params))
372
381
373 # update default_timeout
382 # update default_timeout
374 default_timeout = max(default_timeout, engine_timeout)
383 default_timeout = max(default_timeout, engine_timeout)
searx/webadapter.py+17−19
@@ -1,14 +1,14 @@
1from searx.exceptions import SearxParameterException
1from searx.exceptions import SearxParameterException
2from searx.query import RawTextQuery, VALID_LANGUAGE_CODE
2from searx.query import RawTextQuery, VALID_LANGUAGE_CODE
3from searx.engines import categories, engines
3from searx.engines import categories, engines
4−from searx.search import SearchQuery
4+from searx.search import SearchQuery, EngineRef
5
5
6
6
7# remove duplicate queries.
7# remove duplicate queries.
8# FIXME: does not fix "!music !soundcloud", because the categories are 'none' and 'music'
8# FIXME: does not fix "!music !soundcloud", because the categories are 'none' and 'music'
9−def deduplicate_query_engines(query_engines):
9+def deduplicate_engineref_list(engineref_list):
10− uniq_query_engines = {q["category"] + '|' + q["name"]: q for q in query_engines}
10+ engineref_dict = {q.category + '|' + q.name: q for q in engineref_list}
11− return uniq_query_engines.values()
11+ return engineref_dict.values()
12
12
13
13
14def get_search_query_from_webapp(preferences, form):
14def get_search_query_from_webapp(preferences, form):
@@ -68,7 +68,7 @@ def get_search_query_from_webapp(preferences, form):
68 raise SearxParameterException('time_range', query_time_range)
68 raise SearxParameterException('time_range', query_time_range)
69
69
70 # query_engines
70 # query_engines
71− query_engines = raw_text_query.engines
71+ query_engineref_list = raw_text_query.enginerefs
72
72
73 # timeout_limit
73 # timeout_limit
74 query_timeout = raw_text_query.timeout_limit
74 query_timeout = raw_text_query.timeout_limit
@@ -87,13 +87,13 @@ def get_search_query_from_webapp(preferences, form):
87
87
88 # if engines are calculated from query,
88 # if engines are calculated from query,
89 # set categories by using that informations
89 # set categories by using that informations
90− if query_engines and raw_text_query.specific:
90+ if query_engineref_list and raw_text_query.specific:
91 additional_categories = set()
91 additional_categories = set()
92− for engine in query_engines:
92+ for engineref in query_engineref_list:
93− if 'from_bang' in engine and engine['from_bang']:
93+ if engineref.from_bang:
94 additional_categories.add('none')
94 additional_categories.add('none')
95 else:
95 else:
96− additional_categories.add(engine['category'])
96+ additional_categories.add(engineref.category)
97 query_categories = list(additional_categories)
97 query_categories = list(additional_categories)
98
98
99 # otherwise, using defined categories to
99 # otherwise, using defined categories to
@@ -105,11 +105,10 @@ def get_search_query_from_webapp(preferences, form):
105 if pd_name == 'categories':
105 if pd_name == 'categories':
106 query_categories.extend(categ for categ in map(str.strip, pd.split(',')) if categ in categories)
106 query_categories.extend(categ for categ in map(str.strip, pd.split(',')) if categ in categories)
107 elif pd_name == 'engines':
107 elif pd_name == 'engines':
108− pd_engines = [{'category': engines[engine].categories[0],
108+ pd_engines = [EngineRef(engineref, engines[engineref].categories[0])
109− 'name': engine}
110 for engine in map(str.strip, pd.split(',')) if engine in engines]
109 for engine in map(str.strip, pd.split(',')) if engine in engines]
111 if pd_engines:
110 if pd_engines:
112− query_engines.extend(pd_engines)
111+ query_engineref_list.extend(pd_engines)
113 load_default_categories = False
112 load_default_categories = False
114 elif pd_name.startswith('category_'):
113 elif pd_name.startswith('category_'):
115 category = pd_name[9:]
114 category = pd_name[9:]
@@ -128,7 +127,7 @@ def get_search_query_from_webapp(preferences, form):
128 if not load_default_categories:
127 if not load_default_categories:
129 if not query_categories:
128 if not query_categories:
130 query_categories = list(set(engine['category']
129 query_categories = list(set(engine['category']
131− for engine in query_engines))
130+ for engine in query_engineref_list))
132 else:
131 else:
133 # if no category is specified for this search,
132 # if no category is specified for this search,
134 # using user-defined default-configuration which
133 # using user-defined default-configuration which
@@ -147,15 +146,14 @@ def get_search_query_from_webapp(preferences, form):
147 # using all engines for that search, which are
146 # using all engines for that search, which are
148 # declared under the specific categories
147 # declared under the specific categories
149 for categ in query_categories:
148 for categ in query_categories:
150− query_engines.extend({'category': categ,
149+ query_engineref_list.extend(EngineRef(engine.name, categ)
151− 'name': engine.name}
152− for engine in categories[categ]
150+ for engine in categories[categ]
153− if (engine.name, categ) not in disabled_engines)
151+ if (engine.name, categ) not in disabled_engines)
154
152
155− query_engines = deduplicate_query_engines(query_engines)
153+ query_engineref_list = deduplicate_engineref_list(query_engineref_list)
156 external_bang = raw_text_query.external_bang
154 external_bang = raw_text_query.external_bang
157
155
158− return (SearchQuery(query, query_engines, query_categories,
156+ return (SearchQuery(query, query_engineref_list, query_categories,
159 query_lang, query_safesearch, query_pageno,
157 query_lang, query_safesearch, query_pageno,
160 query_time_range, query_timeout, preferences,
158 query_time_range, query_timeout, preferences,
161 external_bang=external_bang),
159 external_bang=external_bang),
tests/unit/test_search.py+11−11
@@ -5,7 +5,7 @@ from searx.preferences import Preferences
5from searx.engines import engines
5from searx.engines import engines
6
6
7import searx.search
7import searx.search
8−from searx.search import SearchQuery
8+from searx.search import SearchQuery, EngineRef
9
9
10
10
11SAFESEARCH = 0
11SAFESEARCH = 0
@@ -41,7 +41,7 @@ class SearchTestCase(SearxTestCase):
41
41
42 def test_timeout_simple(self):
42 def test_timeout_simple(self):
43 searx.search.max_request_timeout = None
43 searx.search.max_request_timeout = None
44− search_query = SearchQuery('test', [{'category': 'general', 'name': PUBLIC_ENGINE_NAME}],
44+ search_query = SearchQuery('test', [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
45 ['general'], 'en-US', SAFESEARCH, PAGENO, None, None,
45 ['general'], 'en-US', SAFESEARCH, PAGENO, None, None,
46 preferences=Preferences(['oscar'], ['general'], engines, []))
46 preferences=Preferences(['oscar'], ['general'], engines, []))
47 search = searx.search.Search(search_query)
47 search = searx.search.Search(search_query)
@@ -50,7 +50,7 @@ class SearchTestCase(SearxTestCase):
50
50
51 def test_timeout_query_above_default_nomax(self):
51 def test_timeout_query_above_default_nomax(self):
52 searx.search.max_request_timeout = None
52 searx.search.max_request_timeout = None
53− search_query = SearchQuery('test', [{'category': 'general', 'name': PUBLIC_ENGINE_NAME}],
53+ search_query = SearchQuery('test', [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
54 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 5.0,
54 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 5.0,
55 preferences=Preferences(['oscar'], ['general'], engines, []))
55 preferences=Preferences(['oscar'], ['general'], engines, []))
56 search = searx.search.Search(search_query)
56 search = searx.search.Search(search_query)
@@ -59,7 +59,7 @@ class SearchTestCase(SearxTestCase):
59
59
60 def test_timeout_query_below_default_nomax(self):
60 def test_timeout_query_below_default_nomax(self):
61 searx.search.max_request_timeout = None
61 searx.search.max_request_timeout = None
62− search_query = SearchQuery('test', [{'category': 'general', 'name': PUBLIC_ENGINE_NAME}],
62+ search_query = SearchQuery('test', [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
63 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 1.0,
63 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 1.0,
64 preferences=Preferences(['oscar'], ['general'], engines, []))
64 preferences=Preferences(['oscar'], ['general'], engines, []))
65 search = searx.search.Search(search_query)
65 search = searx.search.Search(search_query)
@@ -68,7 +68,7 @@ class SearchTestCase(SearxTestCase):
68
68
69 def test_timeout_query_below_max(self):
69 def test_timeout_query_below_max(self):
70 searx.search.max_request_timeout = 10.0
70 searx.search.max_request_timeout = 10.0
71− search_query = SearchQuery('test', [{'category': 'general', 'name': PUBLIC_ENGINE_NAME}],
71+ search_query = SearchQuery('test', [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
72 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 5.0,
72 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 5.0,
73 preferences=Preferences(['oscar'], ['general'], engines, []))
73 preferences=Preferences(['oscar'], ['general'], engines, []))
74 search = searx.search.Search(search_query)
74 search = searx.search.Search(search_query)
@@ -77,7 +77,7 @@ class SearchTestCase(SearxTestCase):
77
77
78 def test_timeout_query_above_max(self):
78 def test_timeout_query_above_max(self):
79 searx.search.max_request_timeout = 10.0
79 searx.search.max_request_timeout = 10.0
80− search_query = SearchQuery('test', [{'category': 'general', 'name': PUBLIC_ENGINE_NAME}],
80+ search_query = SearchQuery('test', [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
81 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 15.0,
81 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 15.0,
82 preferences=Preferences(['oscar'], ['general'], engines, []))
82 preferences=Preferences(['oscar'], ['general'], engines, []))
83 search = searx.search.Search(search_query)
83 search = searx.search.Search(search_query)
@@ -85,7 +85,7 @@ class SearchTestCase(SearxTestCase):
85 self.assertEqual(search.actual_timeout, 10.0)
85 self.assertEqual(search.actual_timeout, 10.0)
86
86
87 def test_query_private_engine_without_token(self):
87 def test_query_private_engine_without_token(self):
88− search_query = SearchQuery('test', [{'category': 'general', 'name': PRIVATE_ENGINE_NAME}],
88+ search_query = SearchQuery('test', [EngineRef(PRIVATE_ENGINE_NAME, 'general')],
89 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 2.0,
89 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 2.0,
90 preferences=Preferences(['oscar'], ['general'], engines, []))
90 preferences=Preferences(['oscar'], ['general'], engines, []))
91 search = searx.search.Search(search_query)
91 search = searx.search.Search(search_query)
@@ -95,7 +95,7 @@ class SearchTestCase(SearxTestCase):
95 def test_query_private_engine_with_incorrect_token(self):
95 def test_query_private_engine_with_incorrect_token(self):
96 preferences_with_tokens = Preferences(['oscar'], ['general'], engines, [])
96 preferences_with_tokens = Preferences(['oscar'], ['general'], engines, [])
97 preferences_with_tokens.parse_dict({'tokens': 'bad-token'})
97 preferences_with_tokens.parse_dict({'tokens': 'bad-token'})
98− search_query = SearchQuery('test', [{'category': 'general', 'name': PRIVATE_ENGINE_NAME}],
98+ search_query = SearchQuery('test', [EngineRef(PRIVATE_ENGINE_NAME, 'general')],
99 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 2.0,
99 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 2.0,
100 preferences=preferences_with_tokens)
100 preferences=preferences_with_tokens)
101 search = searx.search.Search(search_query)
101 search = searx.search.Search(search_query)
@@ -105,7 +105,7 @@ class SearchTestCase(SearxTestCase):
105 def test_query_private_engine_with_correct_token(self):
105 def test_query_private_engine_with_correct_token(self):
106 preferences_with_tokens = Preferences(['oscar'], ['general'], engines, [])
106 preferences_with_tokens = Preferences(['oscar'], ['general'], engines, [])
107 preferences_with_tokens.parse_dict({'tokens': 'my-token'})
107 preferences_with_tokens.parse_dict({'tokens': 'my-token'})
108− search_query = SearchQuery('test', [{'category': 'general', 'name': PRIVATE_ENGINE_NAME}],
108+ search_query = SearchQuery('test', [EngineRef(PRIVATE_ENGINE_NAME, 'general')],
109 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 2.0,
109 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 2.0,
110 preferences=preferences_with_tokens)
110 preferences=preferences_with_tokens)
111 search = searx.search.Search(search_query)
111 search = searx.search.Search(search_query)
@@ -114,7 +114,7 @@ class SearchTestCase(SearxTestCase):
114
114
115 def test_external_bang(self):
115 def test_external_bang(self):
116 search_query = SearchQuery('yes yes',
116 search_query = SearchQuery('yes yes',
117− [{'category': 'general', 'name': PUBLIC_ENGINE_NAME}],
117+ [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
118 ['general'], 'en-US', SAFESEARCH, PAGENO, None, None,
118 ['general'], 'en-US', SAFESEARCH, PAGENO, None, None,
119 preferences=Preferences(['oscar'], ['general'], engines, [],),
119 preferences=Preferences(['oscar'], ['general'], engines, [],),
120 external_bang="yt")
120 external_bang="yt")
@@ -124,7 +124,7 @@ class SearchTestCase(SearxTestCase):
124 self.assertTrue(results.redirect_url is not None)
124 self.assertTrue(results.redirect_url is not None)
125
125
126 search_query = SearchQuery('youtube never gonna give you up',
126 search_query = SearchQuery('youtube never gonna give you up',
127− [{'category': 'general', 'name': PUBLIC_ENGINE_NAME}],
127+ [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
128 ['general'], 'en-US', SAFESEARCH, PAGENO, None, None,
128 ['general'], 'en-US', SAFESEARCH, PAGENO, None, None,
129 preferences=Preferences(['oscar'], ['general'], engines, []),)
129 preferences=Preferences(['oscar'], ['general'], engines, []),)
130
130