logo

searx

My custom branche(s) on searx, a meta-search engine
commit: f56649e3a9fd6f6b0dd4febe1d2fb377ad0bc528
parent: c7b5cddc485768f47a75b2432a5aebd0130f676c
Author: asciimoo <asciimoo@gmail.com>
Date:   Sun, 20 Oct 2013 11:12:10 +0200

[enh] deviantart engine added

Diffstat:

Asearx/engines/deviantart.py29+++++++++++++++++++++++++++++
1 file changed, 29 insertions(+), 0 deletions(-)

diff --git a/searx/engines/deviantart.py b/searx/engines/deviantart.py @@ -0,0 +1,29 @@ +from urllib import urlencode +from lxml import html +from urlparse import urljoin + +categories = ['images'] + +base_url = 'https://www.deviantart.com/' +search_url = base_url+'search?' + +def request(query, params): + global search_url + params['url'] = search_url + urlencode({'q': query}) + return params + + +def response(resp): + global base_url + results = [] + if resp.status_code == 302: + return results + dom = html.fromstring(resp.text) + for result in dom.xpath('//div[contains(@class, "tt-a tt-fh")]'): + link = result.xpath('.//a[contains(@class, "thumb")]')[0] + url = urljoin(base_url, link.attrib.get('href')) + title_links = result.xpath('.//span[@class="details"]//a[contains(@class, "t")]') + title = ''.join(title_links[0].xpath('.//text()')) + content = html.tostring(link)+'<br />'+link.attrib.get('title') + results.append({'url': url, 'title': title, 'content': content}) + return results