@@ -26,6 +26,7 @@ from searx import settings
2626from searx.utils import gen_useragent
2727import sys
2828from datetime import datetime
29+from flask.ext.babel import gettext
2930
3031engine_dir = dirname(realpath(__file__))
3132
@@ -35,6 +36,7 @@ engines = {}
3536
3637categories = {'general': []}
3738
39+
3840def load_module(filename):
3941 modname = splitext(filename)[0]
4042 if modname in sys.modules:
@@ -50,7 +52,7 @@ if not 'engines' in settings or not settings['engines']:
5052
5153for engine_data in settings['engines']:
5254 engine_name = engine_data['engine']
53− engine = load_module(engine_name+'.py')
55+ engine = load_module(engine_name + '.py')
5456 for param_name in engine_data:
5557 if param_name == 'engine':
5658 continue
@@ -58,38 +60,50 @@ for engine_data in settings['engines']:
5860 if engine_data['categories'] == 'none':
5961 engine.categories = []
6062 else:
61− engine.categories = map(str.strip, engine_data['categories'].split(','))
63+ engine.categories = map(
64+ str.strip, engine_data['categories'].split(','))
6265 continue
6366 setattr(engine, param_name, engine_data[param_name])
6467 for engine_attr in dir(engine):
6568 if engine_attr.startswith('_'):
6669 continue
67− if getattr(engine, engine_attr) == None:
68− print '[E] Engine config error: Missing attribute "{0}.{1}"'.format(engine.name, engine_attr)
70+ if getattr(engine, engine_attr) is None:
71+ print '[E] Engine config error: Missing attribute "{0}.{1}"'.format(engine.name, engine_attr) # noqa
6972 sys.exit(1)
7073 engines[engine.name] = engine
71− engine.stats = {'result_count': 0, 'search_count': 0, 'page_load_time': 0, 'score_count': 0, 'errors': 0}
74+ engine.stats = {
75+ 'result_count': 0,
76+ 'search_count': 0,
77+ 'page_load_time': 0,
78+ 'score_count': 0,
79+ 'errors': 0
80+ }
7281 if hasattr(engine, 'categories'):
7382 for category_name in engine.categories:
7483 categories.setdefault(category_name, []).append(engine)
7584 else:
7685 categories['general'].append(engine)
7786
87+
7888def default_request_params():
79− return {'method': 'GET', 'headers': {}, 'data': {}, 'url': '', 'cookies': {}}
89+ return {
90+ 'method': 'GET', 'headers': {}, 'data': {}, 'url': '', 'cookies': {}}
91+
8092
8193def make_callback(engine_name, results, suggestions, callback, params):
8294 # creating a callback wrapper for the search engine results
8395 def process_callback(response, **kwargs):
8496 cb_res = []
8597 response.search_params = params
86− engines[engine_name].stats['page_load_time'] += (datetime.now() - params['started']).total_seconds()
98+ engines[engine_name].stats['page_load_time'] += \
99+ (datetime.now() - params['started']).total_seconds()
87100 try:
88101 search_results = callback(response)
89102 except Exception, e:
90103 engines[engine_name].stats['errors'] += 1
91104 results[engine_name] = cb_res
92− print '[E] Error with engine "{0}":\n\t{1}'.format(engine_name, str(e))
105+ print '[E] Error with engine "{0}":\n\t{1}'.format(
106+ engine_name, str(e))
93107 return
94108 for result in search_results:
95109 result['engine'] = engine_name
@@ -101,23 +115,25 @@ def make_callback(engine_name, results, suggestions, callback, params):
101115 results[engine_name] = cb_res
102116 return process_callback
103117
118+
104119def score_results(results):
105− flat_res = filter(None, chain.from_iterable(izip_longest(*results.values())))
120+ flat_res = filter(
121+ None, chain.from_iterable(izip_longest(*results.values())))
106122 flat_len = len(flat_res)
107123 engines_len = len(results)
108124 results = []
109125 # deduplication + scoring
110− for i,res in enumerate(flat_res):
126+ for i, res in enumerate(flat_res):
111127 res['parsed_url'] = urlparse(res['url'])
112128 res['engines'] = [res['engine']]
113129 weight = 1.0
114130 if hasattr(engines[res['engine']], 'weight'):
115131 weight = float(engines[res['engine']].weight)
116− score = int((flat_len - i)/engines_len)*weight+1
132+ score = int((flat_len - i) / engines_len) * weight + 1
117133 duplicated = False
118134 for new_res in results:
119− p1 = res['parsed_url'].path[:-1] if res['parsed_url'].path.endswith('/') else res['parsed_url'].path
120− p2 = new_res['parsed_url'].path[:-1] if new_res['parsed_url'].path.endswith('/') else new_res['parsed_url'].path
135+ p1 = res['parsed_url'].path[:-1] if res['parsed_url'].path.endswith('/') else res['parsed_url'].path # noqa
136+ p2 = new_res['parsed_url'].path[:-1] if new_res['parsed_url'].path.endswith('/') else new_res['parsed_url'].path # noqa
121137 if res['parsed_url'].netloc == new_res['parsed_url'].netloc and\
122138 p1 == p2 and\
123139 res['parsed_url'].query == new_res['parsed_url'].query and\
@@ -125,7 +141,7 @@ def score_results(results):
125141 duplicated = new_res
126142 break
127143 if duplicated:
128− if len(res.get('content', '')) > len(duplicated.get('content', '')):
144+ if len(res.get('content', '')) > len(duplicated.get('content', '')): # noqa
129145 duplicated['content'] = res['content']
130146 duplicated['score'] += score
131147 duplicated['engines'].append(res['engine'])
@@ -139,6 +155,7 @@ def score_results(results):
139155 results.append(res)
140156 return sorted(results, key=itemgetter('score'), reverse=True)
141157
158+
142159def search(query, request, selected_engines):
143160 global engines, categories, number_of_searches
144161 requests = []
@@ -160,13 +177,20 @@ def search(query, request, selected_engines):
160177 request_params['started'] = datetime.now()
161178 request_params = engine.request(query, request_params)
162179
163− callback = make_callback(selected_engine['name'], results, suggestions, engine.response, request_params)
164−
165− request_args = dict(headers = request_params['headers']
166− ,hooks = dict(response=callback)
167− ,cookies = request_params['cookies']
168− ,timeout = settings['server']['request_timeout']
169− )
180+ callback = make_callback(
181+ selected_engine['name'],
182+ results,
183+ suggestions,
184+ engine.response,
185+ request_params
186+ )
187+
188+ request_args = dict(
189+ headers=request_params['headers'],
190+ hooks=dict(response=callback),
191+ cookies=request_params['cookies'],
192+ timeout=settings['server']['request_timeout']
193+ )
170194
171195 if request_params['method'] == 'GET':
172196 req = grequests.get
@@ -180,7 +204,7 @@ def search(query, request, selected_engines):
180204
181205 requests.append(req(request_params['url'], **request_args))
182206 grequests.map(requests)
183− for engine_name,engine_results in results.items():
207+ for engine_name, engine_results in results.items():
184208 engines[engine_name].stats['search_count'] += 1
185209 engines[engine_name].stats['result_count'] += len(engine_results)
186210
@@ -192,6 +216,7 @@ def search(query, request, selected_engines):
192216
193217 return results, suggestions
194218
219+
195220def get_engines_stats():
196221 # TODO refactor
197222 pageloads = []
@@ -200,14 +225,15 @@ def get_engines_stats():
200225 errors = []
201226 scores_per_result = []
202227
203− max_pageload = max_results = max_score = max_errors = max_score_per_result = 0
228+ max_pageload = max_results = max_score = max_errors = max_score_per_result = 0 # noqa
204229 for engine in engines.values():
205230 if engine.stats['search_count'] == 0:
206231 continue
207− results_num = engine.stats['result_count']/float(engine.stats['search_count'])
208− load_times = engine.stats['page_load_time']/float(engine.stats['search_count'])
232+ results_num = \
233+ engine.stats['result_count'] / float(engine.stats['search_count'])
234+ load_times = engine.stats['page_load_time'] / float(engine.stats['search_count']) # noqa
209235 if results_num:
210− score = engine.stats['score_count'] / float(engine.stats['search_count'])
236+ score = engine.stats['score_count'] / float(engine.stats['search_count']) # noqa
211237 score_per_result = score / results_num
212238 else:
213239 score = score_per_result = 0.0
@@ -220,30 +246,48 @@ def get_engines_stats():
220246 results.append({'avg': results_num, 'name': engine.name})
221247 scores.append({'avg': score, 'name': engine.name})
222248 errors.append({'avg': engine.stats['errors'], 'name': engine.name})
223− scores_per_result.append({'avg': score_per_result, 'name': engine.name})
249+ scores_per_result.append({
250+ 'avg': score_per_result,
251+ 'name': engine.name
252+ })
224253
225254 for engine in pageloads:
226− engine['percentage'] = int(engine['avg']/max_pageload*100)
255+ engine['percentage'] = int(engine['avg'] / max_pageload * 100)
227256
228257 for engine in results:
229− engine['percentage'] = int(engine['avg']/max_results*100)
258+ engine['percentage'] = int(engine['avg'] / max_results * 100)
230259
231260 for engine in scores:
232− engine['percentage'] = int(engine['avg']/max_score*100)
261+ engine['percentage'] = int(engine['avg'] / max_score * 100)
233262
234263 for engine in scores_per_result:
235− engine['percentage'] = int(engine['avg']/max_score_per_result*100)
264+ engine['percentage'] = int(engine['avg'] / max_score_per_result * 100)
236265
237266 for engine in errors:
238267 if max_errors:
239− engine['percentage'] = int(float(engine['avg'])/max_errors*100)
268+ engine['percentage'] = int(float(engine['avg']) / max_errors * 100)
240269 else:
241270 engine['percentage'] = 0
242271
243−
244− return [('Page loads (sec)', sorted(pageloads, key=itemgetter('avg')))
245− ,('Number of results', sorted(results, key=itemgetter('avg'), reverse=True))
246− ,('Scores', sorted(scores, key=itemgetter('avg'), reverse=True))
247− ,('Scores per result', sorted(scores_per_result, key=itemgetter('avg'), reverse=True))
248− ,('Errors', sorted(errors, key=itemgetter('avg'), reverse=True))
249− ]
272+ return [
273+ (
274+ gettext('Page loads (sec)'),
275+ sorted(pageloads, key=itemgetter('avg'))
276+ ),
277+ (
278+ gettext('Number of results'),
279+ sorted(results, key=itemgetter('avg'), reverse=True)
280+ ),
281+ (
282+ gettext('Scores'),
283+ sorted(scores, key=itemgetter('avg'), reverse=True)
284+ ),
285+ (
286+ gettext('Scores per result'),
287+ sorted(scores_per_result, key=itemgetter('avg'), reverse=True)
288+ ),
289+ (
290+ gettext('Errors'),
291+ sorted(errors, key=itemgetter('avg'), reverse=True)
292+ ),
293+ ]