diff options
| author | Adam Tauber <adam.tauber@balabit.com> | 2015-02-02 09:36:43 +0100 |
|---|---|---|
| committer | Adam Tauber <adam.tauber@balabit.com> | 2015-02-02 09:36:43 +0100 |
| commit | 7f865356f9a6c1b40d0c668c59b3d081de618bac (patch) | |
| tree | 60e9acb27577968a41136c04f248c24871e83860 /searx/engines/www500px.py | |
| parent | 03137eebd9fdfaa57452cb364c1bc9f31b243f67 (diff) | |
| parent | 5a16077455ef9e821a2b5f5f7e975be8a37ce83d (diff) | |
Merge branch 'unit-tests' of https://github.com/Cqoicebordel/searx into Cqoicebordel-unit-tests
Conflicts:
searx/tests/test_engines.py
Diffstat (limited to 'searx/engines/www500px.py')
| -rw-r--r-- | searx/engines/www500px.py | 11 |
1 files changed, 6 insertions, 5 deletions
diff --git a/searx/engines/www500px.py b/searx/engines/www500px.py index f25678c24..99dba4abf 100644 --- a/searx/engines/www500px.py +++ b/searx/engines/www500px.py @@ -15,6 +15,7 @@ from urllib import urlencode from urlparse import urljoin from lxml import html import re +from searx.engines.xpath import extract_text # engine dependent config categories = ['images'] @@ -22,7 +23,7 @@ paging = True # search-url base_url = 'https://500px.com' -search_url = base_url+'/search?search?page={pageno}&type=photos&{query}' +search_url = base_url + '/search?search?page={pageno}&type=photos&{query}' # do search-request @@ -44,11 +45,11 @@ def response(resp): for result in dom.xpath('//div[@class="photo"]'): link = result.xpath('.//a')[0] url = urljoin(base_url, link.attrib.get('href')) - title = result.xpath('.//div[@class="title"]//text()')[0] - thumbnail_src = link.xpath('.//img')[0].attrib['src'] + title = extract_text(result.xpath('.//div[@class="title"]')) + thumbnail_src = link.xpath('.//img')[0].attrib.get('src') # To have a bigger thumbnail, uncomment the next line - #thumbnail_src = regex.sub('4.jpg', thumbnail_src) - content = result.xpath('.//div[@class="info"]//text()')[0] + # thumbnail_src = regex.sub('4.jpg', thumbnail_src) + content = extract_text(result.xpath('.//div[@class="info"]')) img_src = regex.sub('2048.jpg', thumbnail_src) # append result |