summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorasciimoo <asciimoo@gmail.com>2013-10-19 18:29:39 +0200
committerasciimoo <asciimoo@gmail.com>2013-10-19 18:29:39 +0200
commit57eaeb9b74cd73d94863a6b8d4bbb7cd7e455197 (patch)
tree46ed1b692b56b60a0df0562c31ae40d6d4b755c8
parente85972b5e1d296a7eca1eb0d27cdfcab7bdf9360 (diff)
[enh] startpage engine added
-rw-r--r--searx/engines/startpage.py28
1 files changed, 28 insertions, 0 deletions
diff --git a/searx/engines/startpage.py b/searx/engines/startpage.py
new file mode 100644
index 000000000..bad5fa1c3
--- /dev/null
+++ b/searx/engines/startpage.py
@@ -0,0 +1,28 @@
+from urllib import quote
+from lxml import html
+from urlparse import urljoin
+from cgi import escape
+
+base_url = 'http://startpage.com/'
+search_url = base_url+'do/search'
+
+def request(query, params):
+ global search_url
+ query = quote(query.replace(' ', '+'), safe='+')
+ params['url'] = search_url
+ params['method'] = 'POST'
+ params['data'] = {'query': query}
+ return params
+
+
+def response(resp):
+ global base_url
+ results = []
+ dom = html.fromstring(resp.text)
+ for result in dom.xpath('//div[@class="result"]'):
+ link = result.xpath('.//h3/a')[0]
+ url = urljoin(base_url, link.attrib.get('href'))
+ title = ' '.join(link.xpath('.//text()'))
+ content = escape(' '.join(result.xpath('.//p[@class="desc"]//text()')))
+ results.append({'url': url, 'title': title, 'content': content})
+ return results