git.iko.soy/searx
on master help sign in
fix vimeo engine and add comments
58a443bThomas Pointhuber
searx/engines/vimeo.py+40−22
compare with master
unifiedsplit
@@ -1,43 +1,58 @@
1+## Vimeo (Videos)
2+#
3+# @website https://vimeo.com/
4+# @provide-api yes (http://developer.vimeo.com/api), they have a maximum count of queries/hour
5+#
6+# @using-api no (TODO, rewrite to api)
7+# @results HTML (using search portal)
8+# @stable no (HTML can change)
9+# @parse url, title, publishedDate, thumbnail
10+#
11+# @todo rewrite to api
12+# @todo set content-parameter with correct data
13+
1from urllib import urlencode
14from urllib import urlencode
2from HTMLParser import HTMLParser
15from HTMLParser import HTMLParser
3from lxml import html
16from lxml import html
4from searx.engines.xpath import extract_text
17from searx.engines.xpath import extract_text
5from dateutil import parser
18from dateutil import parser
6
19
7−base_url = 'http://vimeo.com'
20+# engine dependent config
8−search_url = base_url + '/search?{query}'
21+categories = ['videos']
9−url_xpath = None
22+paging = True
10−content_xpath = None
11−title_xpath = None
12−results_xpath = ''
13−content_tpl = '<a href="{0}"> <img src="{2}"/> </a>'
14−publishedDate_xpath = './/p[@class="meta"]//attribute::datetime'
15
23
16−# the cookie set by vimeo contains all the following values,
24+# search-url
17−# but only __utma seems to be requiered
25+base_url = 'https://vimeo.com'
18−cookie = {
26+search_url = base_url + '/search/page:{pageno}?{query}'
19− #'vuid':'918282893.1027205400'
27+
20− # 'ab_bs':'%7B%223%22%3A279%7D'
28+# specific xpath variables
21− '__utma': '00000000.000#0000000.0000000000.0000000000.0000000000.0'
29+url_xpath = './a/@href'
22− # '__utmb':'18302654.1.10.1388942090'
30+content_xpath = './a/img/@src'
23− #, '__utmc':'18302654'
31+title_xpath = './a/div[@class="data"]/p[@class="title"]/text()'
24− #, '__utmz':'18#302654.1388942090.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none)' # noqa
32+results_xpath = '//div[@id="browse_content"]/ol/li'
25− #, '__utml':'search'
33+publishedDate_xpath = './/p[@class="meta"]//attribute::datetime'
26−}
27
34
28
35
36+# do search-request
29def request(query, params):
37def request(query, params):
38+ params['url'] = search_url.format(pageno=params['pageno'] ,
30− params['url'] = search_url.format(query=urlencode({'q': query}))
39+ query=urlencode({'q': query}))
31− params['cookies'] = cookie
40+
41+ # TODO required?
42+ params['cookies']['__utma'] = '00000000.000#0000000.0000000000.0000000000.0000000000.0'
43+
32 return params
44 return params
33
45
34
46
47+# get response from search-request
35def response(resp):
48def response(resp):
36 results = []
49 results = []
50+
37 dom = html.fromstring(resp.text)
51 dom = html.fromstring(resp.text)
38
52
39 p = HTMLParser()
53 p = HTMLParser()
40
54
55+ # parse results
41 for result in dom.xpath(results_xpath):
56 for result in dom.xpath(results_xpath):
42 url = base_url + result.xpath(url_xpath)[0]
57 url = base_url + result.xpath(url_xpath)[0]
43 title = p.unescape(extract_text(result.xpath(title_xpath)))
58 title = p.unescape(extract_text(result.xpath(title_xpath)))
@@ -45,10 +60,13 @@ def response(resp):
45 publishedDate = parser.parse(extract_text(
60 publishedDate = parser.parse(extract_text(
46 result.xpath(publishedDate_xpath)[0]))
61 result.xpath(publishedDate_xpath)[0]))
47
62
63+ # append result
48 results.append({'url': url,
64 results.append({'url': url,
49 'title': title,
65 'title': title,
50− 'content': content_tpl.format(url, title, thumbnail),
66+ 'content': '',
51 'template': 'videos.html',
67 'template': 'videos.html',
52 'publishedDate': publishedDate,
68 'publishedDate': publishedDate,
53 'thumbnail': thumbnail})
69 'thumbnail': thumbnail})
70+
71+ # return results
54 return results
72 return results
searx/settings.yml+1−5
@@ -153,11 +153,7 @@ engines:
153
153
154 - name : vimeo
154 - name : vimeo
155 engine : vimeo
155 engine : vimeo
156− categories : videos
156+ locale : en-US
157− results_xpath : //div[@id="browse_content"]/ol/li
158− url_xpath : ./a/@href
159− title_xpath : ./a/div[@class="data"]/p[@class="title"]/text()
160− content_xpath : ./a/img/@src
161 shortcut : vm
157 shortcut : vm
162
158
163locales:
159locales: