logo

searx

My custom branche(s) on searx, a meta-search engine git clone https://hacktivis.me/git/searx.git
commit: ab273e2faaa0ad157443ffe0b09194fad17ad743
parent f2cbefeb54164fb659c5901269d0210ad682861c
Author: Cqoicebordel <Cqoicebordel@users.noreply.github.com>
Date:   Fri, 12 Jun 2015 19:53:38 +0200

Fix Vimeo engine.
Solve #368

Diffstat:

Msearx/engines/vimeo.py12++++++------
Msearx/tests/engines/test_vimeo.py53+++++++++++++++++++++++++++++++++++------------------
2 files changed, 41 insertions(+), 24 deletions(-)

diff --git a/searx/engines/vimeo.py b/searx/engines/vimeo.py @@ -27,11 +27,11 @@ base_url = 'https://vimeo.com' search_url = base_url + '/search/page:{pageno}?{query}' # specific xpath variables -results_xpath = '//div[@id="browse_content"]/ol/li' -url_xpath = './a/@href' -title_xpath = './a/div[@class="data"]/p[@class="title"]' -content_xpath = './a/img/@src' -publishedDate_xpath = './/p[@class="meta"]//attribute::datetime' +results_xpath = '//div[contains(@class,"results_grid")]/ul/li' +url_xpath = './/a/@href' +title_xpath = './/span[@class="title"]' +thumbnail_xpath = './/img[@class="js-clip_thumbnail_image"]/@src' +publishedDate_xpath = './/time/attribute::datetime' embedded_url = '<iframe data-src="//player.vimeo.com/video{videoid}" ' +\ 'width="540" height="304" frameborder="0" ' +\ @@ -58,7 +58,7 @@ def response(resp): videoid = result.xpath(url_xpath)[0] url = base_url + videoid title = p.unescape(extract_text(result.xpath(title_xpath))) - thumbnail = extract_text(result.xpath(content_xpath)[0]) + thumbnail = extract_text(result.xpath(thumbnail_xpath)[0]) publishedDate = parser.parse(extract_text(result.xpath(publishedDate_xpath)[0])) embedded = embedded_url.format(videoid=videoid) diff --git a/searx/tests/engines/test_vimeo.py b/searx/tests/engines/test_vimeo.py @@ -1,3 +1,4 @@ +# -*- coding: utf-8 -*- from collections import defaultdict import mock from searx.engines import vimeo @@ -25,26 +26,42 @@ class TestVimeoEngine(SearxTestCase): self.assertEqual(vimeo.response(response), []) html = """ - <div id="browse_content" class="" data-search-id="696d5f8366914ec4ffec33cf7652de384976d4f4"> - <ol class="js-browse_list clearfix browse browse_videos browse_videos_thumbnails kane" + <div id="browse_content" class="results_grid" data-search-id="696d5f8366914ec4ffec33cf7652de384976d4f4"> + <ul class="js-browse_list clearfix browse browse_videos browse_videos_thumbnails kane" data-stream="c2VhcmNoOjo6ZGVzYzp7InF1ZXJ5IjoidGVzdCJ9"> - <li id="clip_100785455" data-start-page="/search/page:1/sort:relevant/" data-position="1"> - <a href="/videoid" title="Futurama 3d (test shot)"> - <img src="http://image.url.webp" - srcset="http://i.vimeocdn.com/video/482375085_590x332.webp 2x" alt="" - class="thumbnail thumbnail_lg_wide"> - <div class="data"> - <p class="title"> - This is the title - </p> - <p class="meta"> - <time datetime="2014-07-15T04:16:27-04:00" - title="mardi 15 juillet 2014 04:16">Il y a 6 mois</time> - </p> - </div> - </a> + <li data-position="7" data-result-id="clip_79600943"> + <div class="clip_thumbnail"> + <a href="/videoid" class="js-result_url"> + <div class="thumbnail_wrapper"> + <img src="http://image.url.webp" class="js-clip_thumbnail_image"> + <div class="overlay overlay_clip_meta"> + <div class="meta_data_footer"> + <span class="clip_upload_date"> + <time datetime="2013-11-17T08:49:09-05:00" + title="dimanche 17 novembre 2013 08:49">Il y a 1 an</time> + </span> + <span class="clip_likes"> + <img src="https://f.vimeocdn.com/images_v6/svg/heart-icon.svg">2 215 + </span> + <span class="clip_comments"> + <img src="https://f.vimeocdn.com/images_v6/svg/comment-icon.svg">75 + </span> + <span class="overlay meta_data_footer clip_duration">01:12</span> + </div> + </div> + </div> + <span class="title">This is the title</span> + </a> + </div> + <div class="clip_thumbnail_attribution"> + <a href="/fedorshmidt"> + <img src="https://i.vimeocdn.com/portrait/6628061_100x100.jpg" class="avatar"> + <span class="display_name">Fedor Shmidt</span> + </a> + <span class="plays">2,1M lectures</span> + </div> </li> - </ol> + </ul> </div> """ response = mock.Mock(text=html)