searx

commit: c7b5cddc485768f47a75b2432a5aebd0130f676c
parent: 0cfc4b4c8ae0197c8def545b7576448d2b11a928
Author: asciimoo <asciimoo@gmail.com>
Date:   Sun, 20 Oct 2013 10:40:02 +0200

[enh] twitter engine added

Diffstat:
A
searx/engines/twitter.py
26
++++++++++++++++++++++++++

1 file changed, 26 insertions(+), 0 deletions(-)
diff --git a/searx/engines/twitter.py b/searx/engines/twitter.py
@@ -0,0 +1,26 @@
+from urlparse import urljoin
+from urllib import urlencode
+from lxml import html
+
+categories = ['social media']
+
+base_url = 'https://twitter.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 = []
+    dom = html.fromstring(resp.text)
+    for tweet in dom.xpath('//li[@data-item-type="tweet"]'):
+        link = tweet.xpath('.//small[@class="time"]//a')[0]
+        url = urljoin(base_url, link.attrib.get('href'))
+        title = ''.join(tweet.xpath('.//span[@class="username js-action-profile-name"]//text()'))
+        content = ''.join(map(html.tostring, tweet.xpath('.//p[@class="js-tweet-text tweet-text"]//*')))
+        results.append({'url': url, 'title': title, 'content': content})
+    return results