git.iko.soy/searx
on master help sign in
fix yacy-engine and add comments
2cd9801Thomas Pointhuber
searx/engines/yacy.py+67−18
compare with master
unifiedsplit
@@ -1,40 +1,89 @@
1+## Yacy (Web, Images, Videos, Music, Files)
2+#
3+# @website http://yacy.net
4+# @provide-api yes (http://www.yacy-websuche.de/wiki/index.php/Dev:APIyacysearch)
5+#
6+# @using-api yes
7+# @results JSON
8+# @stable yes
9+# @parse (general) url, title, content, publishedDate
10+# @parse (images) url, title, img_src
11+#
12+# @todo parse video, audio and file results
13+
1from json import loads
14from json import loads
2from urllib import urlencode
15from urllib import urlencode
16+from dateutil import parser
17+
18+# engine dependent config
19+categories = ['general', 'images'] #TODO , 'music', 'videos', 'files'
20+paging = True
21+language_support = True
22+number_of_results = 5
23+
24+# search-url
25+base_url = 'http://localhost:8090'
26+search_url = '/yacysearch.json?{query}&startRecord={offset}&maximumRecords={limit}&contentdom={search_type}&resource=global'
3
27
4−url = 'http://localhost:8090'
28+# yacy specific type-definitions
5−search_url = '/yacysearch.json?{query}&maximumRecords=10'
29+search_types = {'general': 'text',
30+ 'images': 'image',
31+ 'files': 'app',
32+ 'music': 'audio',
33+ 'videos': 'video'}
6
34
7
35
36+# do search-request
8def request(query, params):
37def request(query, params):
38+ offset = (params['pageno'] - 1) * number_of_results
39+ search_type = search_types.get(params['category'], '0')
40+
9− params['url'] = url + search_url.format(query=urlencode({'query': query}))
41+ params['url'] = base_url + search_url.format(query=urlencode({'query': query}),
42+ offset=offset,
43+ limit=number_of_results,
44+ search_type=search_type)
45+
46+ # add language tag if specified
47+ if params['language'] != 'all':
48+ params['url'] += '&lr=lang_' + params['language'].split('_')[0]
49+
10 return params
50 return params
11
51
12
52
53+# get response from search-request
13def response(resp):
54def response(resp):
55+ results = []
56+
14 raw_search_results = loads(resp.text)
57 raw_search_results = loads(resp.text)
15
58
59+ # return empty array if there are no results
16 if not raw_search_results:
60 if not raw_search_results:
17 return []
61 return []
18
62
19 search_results = raw_search_results.get('channels', {})[0].get('items', [])
63 search_results = raw_search_results.get('channels', {})[0].get('items', [])
20
64
65+ if resp.search_params['category'] == 'general':
21− results = []
66+ # parse general results
22−
23− for result in search_results:
67+ for result in search_results:
24− tmp_result = {}
68+ publishedDate = parser.parse(result['pubDate'])
25− tmp_result['title'] = result['title']
26− tmp_result['url'] = result['link']
27− tmp_result['content'] = ''
28−
29− if result['description']:
30− tmp_result['content'] += result['description'] + "<br/>"
31
69
32− if result['pubDate']:
70+ # append result
71+ results.append({'url': result['link'],
72+ 'title': result['title'],
33− tmp_result['content'] += result['pubDate'] + "<br/>"
73+ 'content': result['description'],
74+ 'publishedDate': publishedDate})
34
75
35− if result['size'] != '-1':
76+ elif resp.search_params['category'] == 'images':
36− tmp_result['content'] += result['sizename']
77+ # parse image results
78+ for result in search_results:
79+ # append result
80+ results.append({'url': result['url'],
81+ 'title': result['title'],
82+ 'content': '',
83+ 'img_src': result['image'],
84+ 'template': 'images.html'})
37
85
38− results.append(tmp_result)
86+ #TODO parse video, audio and file results
39
87
88+ # return results
40 return results
89 return results
searx/settings.yml+7
@@ -143,6 +143,13 @@ engines:
143 locale : en-US
143 locale : en-US
144 shortcut : vm
144 shortcut : vm
145
145
146+# - name : yacy
147+# engine : yacy
148+# shortcut : ya
149+# base_url : 'http://localhost:8090'
150+# number_of_results : 5
151+# timeout: 3.0
152+
146locales:
153locales:
147 en : English
154 en : English
148 de : Deutsch
155 de : Deutsch