on master help sign in
[mod] add searx.search.EngineRef

was previously a Dict with two or three keys: name, category, from_bang make clear that this is a engine reference (see tests/unit/test_search.py for example) all variables using this class are renamed accordingly.

commit2929495112e9869af24986b27066163572bc558a authorAlexandre Flament <alex@al-f.net> dateTue 22 Sep 2020, 16:22 (6 years ago) parent2dbc0de
4 files changed+58−57
searx/query.py+8−14 searx/search.py+22−13 searx/webadapter.py+17−19 tests/unit/test_search.py+11−11
searx/query.py+8−14
@@ -20,9 +20,8 @@ along with searx. If not, see < http://www.gnu.org/licenses/ >.
2020import re
2121
2222from searx.languages import language_codes
23−from searx.engines import (
24− categories, engines, engine_shortcuts
25−)
23+from searx.engines import categories, engines, engine_shortcuts
24+from searx.search import EngineRef
2625
2726
2827VALID_LANGUAGE_CODE = re.compile(r'^[a-z]{2,3}(-[a-zA-Z]{2})?$')
@@ -40,7 +39,7 @@ class RawTextQuery:
4039 self.disabled_engines = disabled_engines
4140
4241 self.query_parts = []
43− self.engines = []
42+ self.enginerefs = []
4443 self.languages = []
4544 self.timeout_limit = None
4645 self.external_bang = None
@@ -135,26 +134,21 @@ class RawTextQuery:
135134 parse_next = True
136135 engine_name = engine_shortcuts[prefix]
137136 if engine_name in engines:
138− self.engines.append({'category': 'none',
139− 'name': engine_name,
140− 'from_bang': True})
137+ self.enginerefs.append(EngineRef(engine_name, 'none', True))
141138
142139 # check if prefix is equal with engine name
143140 elif prefix in engines:
144141 parse_next = True
145− self.engines.append({'category': 'none',
146− 'name': prefix,
147− 'from_bang': True})
142+ self.enginerefs.append(EngineRef(prefix, 'none', True))
148143
149144 # check if prefix is equal with categorie name
150145 elif prefix in categories:
151146 # using all engines for that search, which
152147 # are declared under that categorie name
153148 parse_next = True
154− self.engines.extend({'category': prefix,
155− 'name': engine.name}
156− for engine in categories[prefix]
157− if (engine.name, prefix) not in self.disabled_engines)
149+ self.enginerefs.extend(EngineRef(engine.name, prefix)
150+ for engine in categories[prefix]
151+ if (engine.name, prefix) not in self.disabled_engines)
158152
159153 if query_part[0] == '!':
160154 self.specific = True
searx/search.py+22−13
@@ -47,13 +47,24 @@ else:
4747 exit(1)
4848
4949
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+
5061class SearchQuery:
5162 """container for all the search parameters (query, language, etc...)"""
5263
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,
5465 timeout_limit=None, preferences=None, external_bang=None):
5566 self.query = query
56− self.engines = engines
67+ self.engineref_list = engineref_list
5768 self.categories = categories
5869 self.lang = lang
5970 self.safesearch = safesearch
@@ -64,7 +75,7 @@ class SearchQuery:
6475 self.external_bang = external_bang
6576
6677 def __str__(self):
67− return self.query + ";" + str(self.engines)
78+ return self.query + ";" + str(self.engineref_list)
6879
6980
7081def send_http_request(engine, request_params):
@@ -318,13 +329,13 @@ class Search:
318329
319330 return True
320331
321− def _get_params(self, selected_engine, user_agent):
322− if selected_engine['name'] not in engines:
332+ def _get_params(self, engineref, user_agent):
333+ if engineref.name not in engines:
323334 return None, None
324335
325− engine = engines[selected_engine['name']]
336+ engine = engines[engineref.name]
326337
327− if not self._is_accepted(selected_engine['name'], engine):
338+ if not self._is_accepted(engineref.name, engine):
328339 return None, None
329340
330341 # set default request parameters
@@ -341,15 +352,13 @@ class Search:
341352 request_params['safesearch'] = self.search_query.safesearch
342353 request_params['time_range'] = self.search_query.time_range
343354
344− request_params['category'] = selected_engine['category']
355+ request_params['category'] = engineref.category
345356 request_params['pageno'] = self.search_query.pageno
346357
347358 return request_params, engine.timeout
348359
349360 # do search-request
350361 def _get_requests(self):
351− global number_of_searches
352−
353362 # init vars
354363 requests = []
355364
@@ -361,14 +370,14 @@ class Search:
361370 default_timeout = 0
362371
363372 # 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:
365374 # 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)
367376 if request_params is None:
368377 continue
369378
370379 # 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))
372381
373382 # update default_timeout
374383 default_timeout = max(default_timeout, engine_timeout)
searx/webadapter.py+17−19
@@ -1,14 +1,14 @@
11from searx.exceptions import SearxParameterException
22from searx.query import RawTextQuery, VALID_LANGUAGE_CODE
33from searx.engines import categories, engines
4−from searx.search import SearchQuery
4+from searx.search import SearchQuery, EngineRef
55
66
77# remove duplicate queries.
88# FIXME: does not fix "!music !soundcloud", because the categories are 'none' and 'music'
9−def deduplicate_query_engines(query_engines):
10− uniq_query_engines = {q["category"] + '|' + q["name"]: q for q in query_engines}
11− return uniq_query_engines.values()
9+def deduplicate_engineref_list(engineref_list):
10+ engineref_dict = {q.category + '|' + q.name: q for q in engineref_list}
11+ return engineref_dict.values()
1212
1313
1414def get_search_query_from_webapp(preferences, form):
@@ -68,7 +68,7 @@ def get_search_query_from_webapp(preferences, form):
6868 raise SearxParameterException('time_range', query_time_range)
6969
7070 # query_engines
71− query_engines = raw_text_query.engines
71+ query_engineref_list = raw_text_query.enginerefs
7272
7373 # timeout_limit
7474 query_timeout = raw_text_query.timeout_limit
@@ -87,13 +87,13 @@ def get_search_query_from_webapp(preferences, form):
8787
8888 # if engines are calculated from query,
8989 # 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:
9191 additional_categories = set()
92− for engine in query_engines:
93− if 'from_bang' in engine and engine['from_bang']:
92+ for engineref in query_engineref_list:
93+ if engineref.from_bang:
9494 additional_categories.add('none')
9595 else:
96− additional_categories.add(engine['category'])
96+ additional_categories.add(engineref.category)
9797 query_categories = list(additional_categories)
9898
9999 # otherwise, using defined categories to
@@ -105,11 +105,10 @@ def get_search_query_from_webapp(preferences, form):
105105 if pd_name == 'categories':
106106 query_categories.extend(categ for categ in map(str.strip, pd.split(',')) if categ in categories)
107107 elif pd_name == 'engines':
108− pd_engines = [{'category': engines[engine].categories[0],
109− 'name': engine}
108+ pd_engines = [EngineRef(engineref, engines[engineref].categories[0])
110109 for engine in map(str.strip, pd.split(',')) if engine in engines]
111110 if pd_engines:
112− query_engines.extend(pd_engines)
111+ query_engineref_list.extend(pd_engines)
113112 load_default_categories = False
114113 elif pd_name.startswith('category_'):
115114 category = pd_name[9:]
@@ -128,7 +127,7 @@ def get_search_query_from_webapp(preferences, form):
128127 if not load_default_categories:
129128 if not query_categories:
130129 query_categories = list(set(engine['category']
131− for engine in query_engines))
130+ for engine in query_engineref_list))
132131 else:
133132 # if no category is specified for this search,
134133 # using user-defined default-configuration which
@@ -147,15 +146,14 @@ def get_search_query_from_webapp(preferences, form):
147146 # using all engines for that search, which are
148147 # declared under the specific categories
149148 for categ in query_categories:
150− query_engines.extend({'category': categ,
151− 'name': engine.name}
152− for engine in categories[categ]
153− if (engine.name, categ) not in disabled_engines)
149+ query_engineref_list.extend(EngineRef(engine.name, categ)
150+ for engine in categories[categ]
151+ if (engine.name, categ) not in disabled_engines)
154152
155− query_engines = deduplicate_query_engines(query_engines)
153+ query_engineref_list = deduplicate_engineref_list(query_engineref_list)
156154 external_bang = raw_text_query.external_bang
157155
158− return (SearchQuery(query, query_engines, query_categories,
156+ return (SearchQuery(query, query_engineref_list, query_categories,
159157 query_lang, query_safesearch, query_pageno,
160158 query_time_range, query_timeout, preferences,
161159 external_bang=external_bang),
tests/unit/test_search.py+11−11
@@ -5,7 +5,7 @@ from searx.preferences import Preferences
55from searx.engines import engines
66
77import searx.search
8−from searx.search import SearchQuery
8+from searx.search import SearchQuery, EngineRef
99
1010
1111SAFESEARCH = 0
@@ -41,7 +41,7 @@ class SearchTestCase(SearxTestCase):
4141
4242 def test_timeout_simple(self):
4343 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')],
4545 ['general'], 'en-US', SAFESEARCH, PAGENO, None, None,
4646 preferences=Preferences(['oscar'], ['general'], engines, []))
4747 search = searx.search.Search(search_query)
@@ -50,7 +50,7 @@ class SearchTestCase(SearxTestCase):
5050
5151 def test_timeout_query_above_default_nomax(self):
5252 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')],
5454 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 5.0,
5555 preferences=Preferences(['oscar'], ['general'], engines, []))
5656 search = searx.search.Search(search_query)
@@ -59,7 +59,7 @@ class SearchTestCase(SearxTestCase):
5959
6060 def test_timeout_query_below_default_nomax(self):
6161 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')],
6363 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 1.0,
6464 preferences=Preferences(['oscar'], ['general'], engines, []))
6565 search = searx.search.Search(search_query)
@@ -68,7 +68,7 @@ class SearchTestCase(SearxTestCase):
6868
6969 def test_timeout_query_below_max(self):
7070 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')],
7272 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 5.0,
7373 preferences=Preferences(['oscar'], ['general'], engines, []))
7474 search = searx.search.Search(search_query)
@@ -77,7 +77,7 @@ class SearchTestCase(SearxTestCase):
7777
7878 def test_timeout_query_above_max(self):
7979 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')],
8181 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 15.0,
8282 preferences=Preferences(['oscar'], ['general'], engines, []))
8383 search = searx.search.Search(search_query)
@@ -85,7 +85,7 @@ class SearchTestCase(SearxTestCase):
8585 self.assertEqual(search.actual_timeout, 10.0)
8686
8787 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')],
8989 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 2.0,
9090 preferences=Preferences(['oscar'], ['general'], engines, []))
9191 search = searx.search.Search(search_query)
@@ -95,7 +95,7 @@ class SearchTestCase(SearxTestCase):
9595 def test_query_private_engine_with_incorrect_token(self):
9696 preferences_with_tokens = Preferences(['oscar'], ['general'], engines, [])
9797 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')],
9999 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 2.0,
100100 preferences=preferences_with_tokens)
101101 search = searx.search.Search(search_query)
@@ -105,7 +105,7 @@ class SearchTestCase(SearxTestCase):
105105 def test_query_private_engine_with_correct_token(self):
106106 preferences_with_tokens = Preferences(['oscar'], ['general'], engines, [])
107107 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')],
109109 ['general'], 'en-US', SAFESEARCH, PAGENO, None, 2.0,
110110 preferences=preferences_with_tokens)
111111 search = searx.search.Search(search_query)
@@ -114,7 +114,7 @@ class SearchTestCase(SearxTestCase):
114114
115115 def test_external_bang(self):
116116 search_query = SearchQuery('yes yes',
117− [{'category': 'general', 'name': PUBLIC_ENGINE_NAME}],
117+ [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
118118 ['general'], 'en-US', SAFESEARCH, PAGENO, None, None,
119119 preferences=Preferences(['oscar'], ['general'], engines, [],),
120120 external_bang="yt")
@@ -124,7 +124,7 @@ class SearchTestCase(SearxTestCase):
124124 self.assertTrue(results.redirect_url is not None)
125125
126126 search_query = SearchQuery('youtube never gonna give you up',
127− [{'category': 'general', 'name': PUBLIC_ENGINE_NAME}],
127+ [EngineRef(PUBLIC_ENGINE_NAME, 'general')],
128128 ['general'], 'en-US', SAFESEARCH, PAGENO, None, None,
129129 preferences=Preferences(['oscar'], ['general'], engines, []),)
130130