@@ -123,6 +123,46 @@ def highlight_content(content, query):
123123
124124 return content
125125
126+def score_results(results):
127+ flat_res = filter(None, chain.from_iterable(izip_longest(*results.values())))
128+ flat_len = len(flat_res)
129+ engines_len = len(results)
130+ results = []
131+ # deduplication + scoring
132+ for i,res in enumerate(flat_res):
133+ res['parsed_url'] = urlparse(res['url'])
134+ res['engines'] = [res['engine']]
135+ weight = 1.0
136+ if hasattr(engines[res['engine']], 'weight'):
137+ weight = float(engines[res['engine']].weight)
138+ elif res['engine'] in settings.weights:
139+ weight = float(settings.weights[res['engine']])
140+ score = int((flat_len - i)/engines_len)*weight+1
141+ duplicated = False
142+ for new_res in results:
143+ p1 = res['parsed_url'].path[:-1] if res['parsed_url'].path.endswith('/') else res['parsed_url'].path
144+ p2 = new_res['parsed_url'].path[:-1] if new_res['parsed_url'].path.endswith('/') else new_res['parsed_url'].path
145+ if res['parsed_url'].netloc == new_res['parsed_url'].netloc and\
146+ p1 == p2 and\
147+ res['parsed_url'].query == new_res['parsed_url'].query and\
148+ res.get('template') == new_res.get('template'):
149+ duplicated = new_res
150+ break
151+ if duplicated:
152+ if len(res.get('content', '')) > len(duplicated.get('content', '')):
153+ duplicated['content'] = res['content']
154+ duplicated['score'] += score
155+ duplicated['engines'].append(res['engine'])
156+ if duplicated['parsed_url'].scheme == 'https':
157+ continue
158+ elif res['parsed_url'].scheme == 'https':
159+ duplicated['url'] = res['parsed_url'].geturl()
160+ duplicated['parsed_url'] = res['parsed_url']
161+ else:
162+ res['score'] = score
163+ results.append(res)
164+ return sorted(results, key=itemgetter('score'), reverse=True)
165+
126166def search(query, request, selected_engines):
127167 global engines, categories, number_of_searches
128168 requests = []
@@ -165,43 +205,8 @@ def search(query, request, selected_engines):
165205 for engine_name,engine_results in results.items():
166206 engines[engine_name].stats['search_count'] += 1
167207 engines[engine_name].stats['result_count'] += len(engine_results)
168− flat_res = filter(None, chain.from_iterable(izip_longest(*results.values())))
169− flat_len = len(flat_res)
170− engines_len = len(selected_engines)
171− results = []
172− # deduplication + scoring
173− for i,res in enumerate(flat_res):
174− res['parsed_url'] = urlparse(res['url'])
175− res['engines'] = [res['engine']]
176− weight = 1.0
177− if hasattr(engines[res['engine']], 'weight'):
178− weight = float(engines[res['engine']].weight)
179− elif res['engine'] in settings.weights:
180− weight = float(settings.weights[res['engine']])
181− score = int((flat_len - i)/engines_len)*weight+1
182− duplicated = False
183− for new_res in results:
184− p1 = res['parsed_url'].path[:-1] if res['parsed_url'].path.endswith('/') else res['parsed_url'].path
185− p2 = new_res['parsed_url'].path[:-1] if new_res['parsed_url'].path.endswith('/') else new_res['parsed_url'].path
186− if res['parsed_url'].netloc == new_res['parsed_url'].netloc and\
187− p1 == p2 and\
188− res['parsed_url'].query == new_res['parsed_url'].query and\
189− res.get('template') == new_res.get('template'):
190− duplicated = new_res
191− break
192− if duplicated:
193− if len(res.get('content', '')) > len(duplicated.get('content', '')):
194− duplicated['content'] = res['content']
195− duplicated['score'] += score
196− duplicated['engines'].append(res['engine'])
197− if duplicated['parsed_url'].scheme == 'https':
198− continue
199− elif res['parsed_url'].scheme == 'https':
200− duplicated['url'] = res['parsed_url'].geturl()
201− duplicated['parsed_url'] = res['parsed_url']
202− else:
203− res['score'] = score
204− results.append(res)
208+
209+ results = score_results(results)
205210
206211 for result in results:
207212 if 'content' in result:
@@ -209,7 +214,7 @@ def search(query, request, selected_engines):
209214 for res_engine in result['engines']:
210215 engines[result['engine']].stats['score_count'] += result['score']
211216
212− return sorted(results, key=itemgetter('score'), reverse=True)
217+ return results
213218
214219def get_engines_stats():
215220 pageloads = []