on master help sign in
[enh] result ordering and deduplication
fa9c9e0asciimoo
searx/engines/__init__.py+21−1
@@ -4,6 +4,7 @@ from os import listdir
4from imp import load_source
4from imp import load_source
5import grequests
5import grequests
6from itertools import izip_longest, chain
6from itertools import izip_longest, chain
7+from operator import itemgetter
7
8
8engine_dir = dirname(realpath(__file__))
9engine_dir = dirname(realpath(__file__))
9
10
@@ -56,4 +57,23 @@ def search(query, request, selected_engines):
56 )
57 )
57 requests.append(req)
58 requests.append(req)
58 grequests.map(requests)
59 grequests.map(requests)
59− return list(filter(None, chain(*izip_longest(*results.values()))))
60+ flat_res = list(filter(None, chain(*izip_longest(*results.values()))))
61+ flat_len = len(flat_res)
62+ results = []
63+ # deduplication + scoring
64+ for i,res in enumerate(flat_res):
65+ score = flat_len - i
66+ duplicated = False
67+ for new_res in results:
68+ if res['url'] == new_res['url']:
69+ duplicated = new_res
70+ break
71+ if duplicated:
72+ if len(res['content']) > len(duplicated):
73+ duplicated['content'] = res['content']
74+ duplicated['score'] += score
75+ else:
76+ res['score'] = score
77+ results.append(res)
78+
79+ return sorted(results, key=itemgetter('score'), reverse=True)